[Senate Hearing 118-308]
[From the U.S. Government Publishing Office]
S. Hrg. 118-308
U.S. LEADERSHIP ON ARTIFICIAL INTELLIGENCE
IN AN ERA OF STRATEGIC COMPETITION
=======================================================================
HEARING
BEFORE THE
COMMITTEE ON FOREIGN RELATIONS
UNITED STATES SENATE
ONE HUNDRED EIGHTEENTH CONGRESS
FIRST SESSION
__________
NOVEMBER 15, 2023
__________
Printed for the use of the Committee on Foreign Relations
[GRAPHIC NOT AVAILABLE IN TIFF FORMAT]
Available via http://www.govinfo.gov
__________
U.S. GOVERNMENT PUBLISHING OFFICE
55-804 PDF WASHINGTON : 2024
-----------------------------------------------------------------------------------
COMMITTEE ON FOREIGN RELATIONS
BENJAMIN L. CARDIN, Maryland, Chairman
ROBERT MENENDEZ, New Jersey JAMES E. RISCH, Idaho
JEANNE SHAHEEN, New Hampshire MARCO RUBIO, Florida
CHRISTOPHER A. COONS, Delaware MITT ROMNEY, Utah
CHRISTOPHER MURPHY, Connecticut PETE RICKETTS, Nebraska
TIM KAINE, Virginia RAND PAUL, Kentucky
JEFF MERKLEY, Oregon TODD YOUNG, Indiana
CORY A. BOOKER, New Jersey JOHN BARRASSO, Wyoming
BRIAN SCHATZ, Hawaii TED CRUZ, Texas
CHRIS VAN HOLLEN, Maryland BILL HAGERTY, Tennessee
TAMMY DUCKWORTH, Illinois TIM SCOTT, South Carolina
Damian Murphy, Staff Director
Christopher M. Socha, Republican Staff Director
John Dutton, Chief Clerk
(ii)
C O N T E N T S
----------
Page
Cardin, Hon. Benjamin L., U.S. Senator From Maryland............. 1
Risch, Hon. James E., U.S. Senator From Idaho.................... 3
Fick, Hon. Nathaniel, Ambassador-at-Large, Bureau for Cyberspace
and
Digital Policy, United States Department of State, Washington,
DC............................................................. 5
Prepared Statement........................................... 7
Graviss, Dr. Matthew, Chief Data and Artificial Intelligence
Officer, United States Department of State, Washington, DC..... 9
Prepared Statement........................................... 11
Additional Material Submitted for the Record
Responses of Mr. Nathaniel Fick to Questions Submitted by Senator
Benjamin L. Cardin............................................. 30
Responses of Dr. Matthew Graviss to Questions Submitted by
Senator
Benjamin L. Cardin............................................. 34
Responses of Mr. Nathaniel Fick to Questions Submitted by Senator
Christopher A. Coons........................................... 36
Responses of Mr. Nathaniel Fick to Questions Submitted by Senator
Brian Schatz................................................... 37
Letter to President Biden, Dated November 30, 2023, Concerning
the Decision of the USTR To Stop Supporting Key Commitments in
the E-Commerce Negotiations at the WTO......................... 39
(iii)
U.S. LEADERSHIP ON ARTIFICIAL INTELLIGENCE IN AN ERA OF STRATEGIC
COMPETITION
----------
WEDNESDAY, NOVEMBER 15, 2023
U.S. Senate,
Committee on Foreign Relations,
Washington, DC.
The committee met, pursuant to notice, at 10:04 a.m., in
room SD-419, Dirksen Senate Office Building, Hon. Benjamin J.
Cardin presiding.
Present: Senators Cardin [presiding], Menendez, Shaheen,
Kaine, Merkley, Van Hollen, Risch, Romney, Ricketts, Young, and
Barrasso.
OPENING STATEMENT OF HON. BENJAMIN L. CARDIN,
U.S. SENATOR FROM MARYLAND
The Chairman. The Senate Foreign Relations Committee will
come to order.
This hearing of the Senate Foreign Relations Committee, the
subject is the AI revolution and I look forward to our
testimony from our witnesses.
As I think most members know, our leadership, including
with Senator Young, have been busy setting up a college for us
to learn AI. We appreciate it very much. We are getting
continuing Senate credits for our attendance at the AI
conferences, so thank you, Senator Young, for your leadership
in putting that together.
This is just another opportunity for the role that the
Foreign Relations Committee will play in regards to what is our
appropriate policies as it relates to AI.
The AI revolution is going to change economics. It is going
to change societies. It is going to change the entire world.
That means it is also going to change the way we do diplomacy.
Today's advanced AI models are 5 billion times more
powerful than just a decade ago--that is 5 billion times--the
kind of technology growth is unlike anything humans have ever
invented, and we are still just at the early stages of the AI
revolution.
It has the potential to usher in unpredictable, complicated
challenges like empowering digital authoritarianism or
spreading bias and disinformation or locking us into a
spiraling AI race against our adversaries.
At the same time, the AI revolution will bring about
positive possibilities unseen in human history--predicting the
outbreak of war, improving harvests to solve food security,
curing deadly diseases, unlocking green energy.
I want to thank our witnesses for appearing before us
today. I know you and your team face challenges. Preparing the
world to responsibly harness and deploy AI will be difficult.
Preparing the Department of State for the future will not be
easy either.
I want to thank you for the accomplishments you have
advanced so far and for the heroic efforts that you are
engaging in. Your dedication and commitment in serving our
nation is appreciated.
Your teams include some of the sharpest minds in our
government. That is why I want to challenge you to do even
more.
Thus far much of the discussions around AI have centered
around the military and private sector, but making money or
making war is too narrow of an approach for the AI revolution.
We need to proceed with an AI agenda that is wrapped in
American values. That is why I think the State Department is
not only crucial to this effort; the State Department must be a
leader in global AI governance efforts.
The State Department has valuable data. That data has not
always been well-organized or used to its fullest potential.
This includes real-time information from embassies and
diplomats stationed around the world. It includes the U.N.
voting records. It includes environmental and economic trends.
This is highly valuable information. We know how much we
are trying to gather information today and how valuable that is
for the private sector. State needs to unlock the insights from
this novel data and you have the workforce to do just that.
The Department has concentrated public servants who have--
with advanced degrees and go beyond foreign policy. Your data
science experts are going to be vital in this effort.
Our diplomats also must be involved as we manage our AI
competition with nations like China and Russia. Diplomats know
how to take on complex and multifaceted problems.
They know how to balance competing ideas when making policy
and, most importantly, American diplomats know how to drive an
agenda based on American values--values that need to be part of
the AI revolution, values that protect people's privacy, values
that lead to democratic elections, respect for human rights,
and peace and security for people around the world.
If we are going to lead on global AI governance, if we are
going to avoid a dangerous AI arms race, if we are going to
harness AI to improve the lives of people on this planet, the
State Department must be in the lead. It cannot be left to
generals without diplomats.
I believe Congress and this committee in particular must
play a central role in guiding America's AI efforts. I hope you
will share your thoughts on how you think Congress can
accomplish that.
I now recognize our distinguished ranking member: Senator
Risch.
STATEMENT OF HON. JAMES E. RISCH,
U.S. SENATOR FROM IDAHO
Senator Risch. Thank you very much, Mr. Chairman.
The ongoing wars in Europe and the Middle East are
reminders of the threats and instability that define the era of
strategic competition in which we live.
Amidst the images of chaos and destruction splashed across
the headlines, yet another competition is arising. New advances
in AI will transform the way we live, do business, and interact
with the world. These advances also have significant
implications for our foreign policy and, importantly, national
security.
The United States and our allies must lead the world in
developing the transformational technologies and the standards
that govern them. That will shape the future. If harnessed
appropriately, AI-driven algorithms can provide the State
Department with real-time data and insights.
This includes everything from how effective policies are in
different parts of the world to which IT vulnerabilities are
most likely to be exploited by an adversary to which overseas
staffing models are most efficient and effective in meeting the
Department's needs.
We must also prepare for the ways our adversaries like
Russia, Iran, and particularly China are trying to use AI. AI-
powered cyber attacks could overwhelm our defenses by rapidly
identifying our vulnerabilities and exploiting them. AI-driven
information operations could target government officials or
certain segments of the public and confuse, distract, or
paralyze the decision-making process.
AI-enabled battlefield management systems could give our
adversaries a decisive advantage in targeting U.S. forces and
striking key weak points. Since AI and machine learning rely on
accurate data to work correctly, our adversaries could also
deliberately manipulate or corrupt publicly available databases
to ensure predictive models or analytics used by the U.S.
Government do not work properly during a crisis.
To respond to both the opportunities and the threats posed
by new AI-driven technologies, the State Department should
focus on two areas, and I hope our witnesses will be able to
talk more about those today.
First, we need to work with our allies to set the standards
that will govern how AI is used around the world. We need to
collaborate on research in key areas and identify the specific
national security-sensitive technologies that adversaries like
China will try to steal or copy and thereafter develop tougher
safeguards to answer those issues.
Ambassador Fick, Congress established the Bureau of
Cyberspace and Digital Policy, which you lead, in part to
spearhead this type of engagement with allies. I look forward
to hearing more from you about the work you are doing in this
area.
I would also like to hear your insights into how AI could
transform the cyber threat landscape and what you are doing
about that.
Second, we need a workforce that not only understands how
data and AI and machine learning work, but also how to
integrate these tools into their daily work.
I am pleased to see the Department begin to pursue a data-
driven approach to diplomacy because it has the potential to
improve our foreign policy. Using all the information available
to make national security decisions is a crucial part in
getting to the right answer.
Data-driven diplomacy also helps counteract groupthink
within the Department which limits options and stifles critical
thinking, but data-driven diplomacy cannot just be a talking
point. The Department has to truly commit to integrating data
into the policymaking process and changing course if necessary
when it receives objective feedback that a policy or procedure
simply is not working.
The State Department's Center for Data Analytics was
established to better integrate data analysis and expertise
into foreign policy decision-making and to develop a workforce
that possesses the skills needed to take advantage of these
technologies.
Dr. Graviss, I look forward to hearing more about the
progress the Center for Analytics has made in expanding data
access and utilization throughout the Department. Any insights
you have into challenges and obstacles that remain would be
greatly appreciated.
Thank you, Mr. Chairman.
The Chairman. Thank you, Senator Risch, which is most of
the cases our opening statements are very comparable and very
similar. Thank you very much for your leadership here and I
look forward to working with you.
Senator Risch. You are not suggesting to plagiarize, are
you?
The Chairman. No, I am suggesting we might want to combine
some of our staff and save some--no, I am just----
[Laughter.]
The Chairman. It is always interesting to hear the opening
statement from my colleague.
We have two very distinguished witnesses today: Ambassador
Fick and Dr. Graviss. I am going to introduce both of you. Your
full statements will be made part of our record.
We would ask that you proceed in about 5 minutes so we have
time for the members to ask questions and to have a discussion.
First, let me introduce Ambassador Fick, who currently
leads the State Department's Bureau of Cyberspace and Digital
Policy. As the inaugural ambassador-at-large, Ambassador Fick
has taken the reins at a critical moment for strengthening U.S.
diplomatic efforts on cyber and digital issues, particularly
the rapid development of AI technologies.
Prior to his leadership of the CDP bureau, Ambassador Fick
had extensive career in the private sector as a technology
executive and entrepreneur.
He has also served as a Marine Corps infantry
reconnaissance officer including combat tours in Afghanistan
and Iraq. We thank you for your service to our country in so
many different ways.
Dr. Graviss has been the Department's chief data officer
since December 2020. He has recently selected to serve as the
Department's inaugural chief artificial intelligence officer in
a dual-hatted capacity.
This also comes at a critical moment as the executive
branch agencies react to the development and deployment of
artificial intelligence and other data analytical tools.
Prior to his leadership of the Department's Center for
Analytics, Dr. Graviss spent 13 years in leadership positions
at the Department of Homeland Security as well as his time in
the private sector.
He holds a master's degree in mechanical engineering from
Texas A&M University and a doctorate in systems engineering
from George Washington University.
We will start with Ambassador Fick.
STATEMENT OF HON. NATHANIEL FICK, AMBASSADOR-AT-LARGE, BUREAU
FOR CYBERSPACE AND DIGITAL POLICY, UNITED STATES DEPARTMENT OF
STATE, WASHINGTON, DC
Mr. Fick. Good morning, Chairman Cardin, Ranking Member
Risch, distinguished members of the committee.
On behalf of the Department of State, thank you for the
opportunity to speak here today and, more broadly, thank you
for your support over the 14 months I have been in this new
role.
I am pleased to provide you today with an update on our
work in the service of the United States international
technology policy priorities and will speak specifically to our
efforts to strengthen U.S. global leadership to unlock the
potentials of AI while also addressing its challenges.
At the State Department, I oversee the organizations that
lead and coordinate the Department's work on cyberspace,
digital policy, and emerging technologies, and there are two of
them: the Bureau of Cyberspace and Digital Policy, known as
CDP, and also the Office of the Special Envoy for Critical and
Emerging Tech, S/TECH for short.
Building on years of bipartisan work, Secretary of State
Blinken established these organizations in partnership with
Congress as part of a comprehensive effort to modernize
American diplomacy and make technology central to U.S. foreign
policy.
He gave us a very clear mission: to shape the terms of the
technology future, and to extend America's industrial and
innovation strategy into the international realm.
Together, CDP and S/TECH with our interagency partners work
to advance U.S. global leadership on topics such as: trusted
technologies in digital infrastructure, advanced wireless
networks, quantum computing, biotechnology and synthetic
biology, cloud services and data centers, undersea
communications cables, satellite communications, and, yes, AI.
We also work to build partner cybersecurity capacity, to
strengthen international technical standards, to advance global
cyber stability and counter adversaries in cyberspace and,
finally, we use technology to promote and protect human rights.
An important distinction between my work and the work of my
colleague appearing with me today, Dr. Graviss, is that my team
focuses on leading U.S. diplomacy on technology topics with
external international audience while Dr. Graviss' work is
focused on AI deployment within the Department.
As someone who has spent a decade building and leading a
software business and after a couple combat tours in the
Marines, I do believe very fundamentally that responsible
technology innovation is increasingly a foundational source of
our geopolitical power and over the past year I have seen
firsthand how these technologies are transforming every aspect
of our societies from how we work to how we learn to how we
communicate, and they are transforming every aspect of our
foreign policy.
Many traditional measures of national strength such as GDP
or military capacity are increasingly downstream of a nation's
ability to innovate in these key technology areas.
Put simply, in geopolitical competition terms, tech is
increasingly the game. It is revolutionary, it is accelerating,
and the United States must lead, engaging boldly on behalf of
our values and our interests.
As the President recently said, we stand at an inflection
point in history. Nowhere is that more evident than with AI.
The decisions that we make today about how we shape the
parameters around AI will define the world in so many ways for
decades to come.
It is a responsibility that we must assume together with
our private sector, with civil society, and with our
international partners. American companies lead the world in AI
innovation and the United States has increasingly strong
momentum domestically to lead in AI governance.
The recent executive order on AI directs the most
significant action any government anywhere in the world has
ever taken on AI safety, security, and trust and it is the
natural next step after having first secured the voluntary
commitments that were agreed to by leading AI companies, and
the President is committed to working with Congress to craft
bipartisan legislation to establish an enduring domestic policy
framework.
At the same time, we are working internationally with our
closest partners to advance our shared views on AI policy. The
G-7, with strong U.S. leadership and engagement, just released
an international code of conduct for AI developers which draws
heavily from the voluntary commitments.
I attended the U.K.'s AI safety summit 2 weeks ago at
Bletchley Park, hallowed ground in the history of technology
and national security, and we continue to work to advance
shared approaches to AI with European partners through
mechanisms including the U.S.-EU Trade and Technology Council.
Countries around the world realize that AI is now the
foundational technology driving advances that can help solve
some of our most pressing shared challenges in science, in
energy, in agriculture, in education and more, and my team's
role is to engage with the world in these conversations to
synchronize multi-stakeholder approaches on governance, to
harness the benefits of AI to mitigate its risks, and to do so
while always ensuring that our innovation ecosystem remains our
North Star.
We do this with three orienting principles in mind. First,
it is essential for the United States to lead with an
affirmative vision for the role of technology in our shared
future.
We believe in tech's power to accelerate innovation, to
solve major global challenges, and to support our future
prosperity. Our affirmative vision is reflected in both our
engagements and also in our foreign assistance through programs
such as AI Connect and other recently announced programs to
leverage AI to help reduce poverty, to address energy needs,
and to improve global public health.
By committing resources to these programs, we were able to
secure follow-on investments from the private sector that
leveraged our commitment of $15 million--leveraged that
commitment several times over to accelerate progress on the
Sustainable Development Goals using AI.
The U.S. is the world's leader in innovation and if we do
not champion the benefits of technology, who will? That is
principle one.
Second, we know we must do this work in partnership with
others. International partnerships on everything from R&D
investment to standards harmonization to regulatory
interoperability are the policy underpinnings that are
necessary in order to seize this moment. No one can do this
alone--no one country regardless of how big or powerful, no one
company regardless of how advanced.
Over the past year, the State Department has been engaged
constantly with our allies and with our partners, bilaterally
and also in multilateral venues, like the G-7, the G-20, the
OECD, the United Nations, to achieve tangible commitments on
the responsible development, deployment, and use of AI, and
through the Executive Order the U.S. is leading by example.
We are providing a model for the responsible use of this
critical technology in our own government and we are committed
to increasing participation in the AI voluntary commitments
among leading companies both in the United States and around
the world.
Third, we recognize that all of this is happening while a
determined and well-resourced set of adversaries and
competitors tirelessly advance a very different view of
technology's role in our future.
They are aggressively trying to reshape the international
rules-based order and are investing heavily in a worldview that
prioritizes authoritarian tactics over democratic values and
governance. Technology governance and specifically AI
governance is one of the geopolitical imperatives of our time.
We need to lead international conversations to shape the
global technology landscape in the future so that AI advances
democratic values and human rights, protects our safety and our
security, and supports our prosperity.
When the United States pulls back, our adversaries fill
that void. Technology with global impact requires global
action. The State Department is delivering on the important
work of technology diplomacy for the American people.
Thank you again for the opportunity to join you today. I
look forward to your questions and your input.
[The prepared statement of Mr. Fick follows:]
Prepared Statement of Mr. Nathaniel Fick
Chairman Cardin, Ranking Member Risch, and distinguished members of
the Committee, on behalf of the Department of State, thank you for the
opportunity to speak here today. More broadly, thank you for your
unfailing support throughout the 14 months I have been in this new
role. I am honored to provide you today with an update on our work in
the service of the United States' international technology policy
priorities and will speak specifically to our efforts to strengthen
U.S. global leadership to unlock the potential of artificial
intelligence (AI) while also addressing its challenges.
At the State Department, I oversee the organizations that lead and
coordinate the Department's work on cyberspace, digital policy, and
emerging technologies: the Bureau of Cyberspace and Digital Policy,
known as CDP, and the Office of the Special Envoy for Critical and
Emerging Technology, S/TECH for short. Building on years of bipartisan
work, Secretary of State Antony Blinken established these organizations
in partnership with Congress as part of a comprehensive effort to
modernize American diplomacy and make technology central to U.S.
foreign policy. He gave us a mission: to shape the terms of the
technology future and to extend the Biden administration's modern
industrial and innovation strategy into the international realm.
Together, CDP and S/TECH, with our interagency partners, work to
advance U.S. leadership globally on topics such as: trusted
technologies and digital infrastructure, artificial intelligence,
advanced wireless networks, quantum computing, biotechnology, cloud
services and data centers, undersea telecommunications cables,
satellite communications; and trusted data flows across borders. We
also work to build partner cybersecurity capacity; strengthen
consensus-based international technical standards; advance global cyber
stability and counter adversaries in cyberspace through sustained
diplomatic initiatives and international partnerships; and finally, we
use technology to promote and protect human rights.
An important distinction between my work and the work of my
colleague appearing with me today, Dr. Matthew Graviss, is that my team
focuses on leading U.S. diplomacy on technology topics with external,
international audiences. So, while my team engages governments around
the world, multilateral institutions, civil society, and industry,
these efforts are stronger when we lead by example and responsibly and
effectively use AI inside the State Department as well. That is where
Dr. Graviss and his team's exciting work comes in, as he will explain
shortly.
As someone who spent a decade building and leading a cybersecurity
software company after a couple of combat tours in the Marines, I
believe that responsible technological innovation is increasingly a
foundational source of geopolitical power. Over the past year, I have
seen firsthand in how technologies are transforming every aspect of our
societies and our economies--how we work, how we learn, how we
communicate, and how we care for ourselves. They are also transforming
every aspect of our foreign policy. Many traditional measures of
strength, such as GDP or military capacity, are increasingly downstream
from our ability to innovate in core technology areas. In that sense,
technology innovation is driving more and more of what is, and is not,
possible in our foreign and national security policy. Put simply, in
the realm of geopolitical competition, tech is the game. It is
revolutionary and it is accelerating. The United States must lead,
engaging boldly on behalf of our values and interests.
As the President recently said, ``we stand at an inflection point
in history'' and nowhere is that more evident than with artificial
intelligence. The decisions that we make collectively today about how
we shape the parameters around AI will define our world for decades. It
is a responsibility that we must assume together with our private
sector, civil society, and our international partners. American
companies lead the world in AI innovation, and the United States has
increasingly strong momentum domestically to lead in AI governance. The
recent Executive Order on AI directs the most significant action any
government anywhere in the world has ever taken on AI safety, security,
and trust. It is just the next step after securing the voluntary
commitments agreed to by leading AI companies, and the President is
committed to working with Congress to craft bipartisan legislation that
establishes an enduring domestic policy framework.
At the same time, we are working internationally with our closest
partners to advance our shared views on AI policy. The G7, with strong
U.S. engagement, just released an International Code of Conduct for AI
developers which draws heavily from the voluntary commitments. I
attended the UK's AI Safety Summit 2 weeks ago at Bletchley Park,
hallowed ground in the history of technology and national security. And
we continue to work to advance shared approaches to AI with our
European partners through the U.S.-EU Trade and Technology Council.
Countries around the world realize that AI is now the foundational
technology driving advances that can help solve some of our most
pressing shared challenges in science, health care, energy,
transportation, education, and more. And AI is already ubiquitous--the
choice is not whether AI will change our societies and economies, but
how governments and societies use it responsibly. We must ensure that
rights-respecting states sustain their competitive advantage, that the
technologies benefit all our people, and that risks are mitigated.
My team's role--in partnership with the U.S. Agency for
International Development, which is leading on how AI can be used to
advance global development--is to engage with the world in these
conversations--to synchronize multi-stakeholder approaches on AI
governance, to harness the benefits of AI, to mitigate its risks, and
to do so while encouraging our innovation ecosystem. We do this with
three orienting principles always in mind:
First, it is essential for the United States to lead with an
affirmative vision for the role of technology in our shared future. We
believe in technology's power to accelerate innovation, to solve major
global challenges, and to support our future prosperity. Our
affirmative vision is reflected in both our engagements and foreign
assistance, through programs such as AI Connect at State and the
Responsible Computing Challenge and Equitable AI Challenge at USAID,
and other recently announced programs to leverage AI to help reduce
poverty, address energy needs, improve global public health, access to
education, and build a stronger coalition around inclusive and ethical
AI governance. This affirmative vision must include a commitment to
elevating the voices of those around the world who are often not
included in the conversations around AI development, deployment, and
the international standards governing AI. By committing resources to
these programs, we were able to secure follow-on investments from the
private sector that didn't just match but far exceeded our commitment
of $15 million to leverage AI to accelerate progress on the Sustainable
Development Goals. The United States is the world's leader in
innovation--if we don't champion the benefits of technology, who will?
Second, we know we must do this work in partnership with others.
International partnerships on everything from R&D investment to
standards harmonization to regulatory interoperability are the policy
underpinnings necessary to seize this moment. No one can do this alone.
Over the past year, the State Department has been engaged constantly
with allies and partners bilaterally and in multilateral venues like
the G7, the G20, the OECD, and the United Nations to achieve tangible
commitments on the responsible development, deployment, and use of
artificial intelligence. Through the AI Executive Order, the United
States is leading by example, providing a model for the responsible use
of this critical technology. We are working to increase participation
in the AI Voluntary Commitments among leading companies in the United
States and all around the world, to broaden adherence to the Code of
Conduct, and to internationalize different elements of the E.O. as a
more comprehensive policy framework. In addition to working the
governments and companies, we--from the President on down--are working
with civil society experts and organizations at every step on this
journey to effectively manage risks related to AI technologies.
Third, we recognize that all this is happening while a determined
and well-resourced set of adversaries and competitors tirelessly
advance a very different view of technology's role in our future. They
are aggressively trying to re-shape the international rules-based order
and are investing heavily in a worldview that prioritizes authoritarian
tactics over democratic values and governance. Technology governance,
and specifically AI, is one of the geopolitical imperatives of our
time. We need to lead international conversations to shape the global
technology landscape of the future, so that AI advances democratic
values and human rights, protects our safety and security, and supports
consumers and workers. When the United States pulls back, our
adversaries and competitors fill the void.
Technology with global impact requires global action, and the State
Department is delivering on the important work of technology diplomacy
for the American people.
Thank you again for the opportunity to join you today. I look
forward to your questions and perspectives.
The Chairman. Thank you very much, Mr. Ambassador.
Dr. Graviss.
STATEMENT OF DR. MATTHEW GRAVISS, CHIEF DATA AND ARTIFICIAL
INTELLIGENCE OFFICER, UNITED STATES DEPARTMENT OF STATE,
WASHINGTON, DC
Dr. Graviss. Chairman Cardin, Ranking Member Risch,
distinguished members of the committee, thank you for this
opportunity.
I would like to echo Ambassador Fick's message that we
appreciate the committee's effort to understand the role of
artificial intelligence both at the Department and in our
global diplomatic efforts.
Serving as the Department's first chief data and AI
officer, I lead the Center for Analytics. Formally launched in
2020, the Center for Analytics is innovating with, among other
things, AI pilots while instilling a culture of responsible AI
use across the Department.
To elevate our collective data and AI proficiency our
stellar team of data scientists and engineers and analysts
partner closely with dozens of bureaus, offices, and posts
across the Department.
Our work delivers on the President's Executive Order on
safe, secure, and trustworthy AI which emphasizes the
transformative potential of AI in government. Just last week,
the Secretary released the Department's first ever enterprise
AI strategy.
This strategy will provide our world-class diplomatic corps
with advanced AI governance and analytics, training and
infrastructure to prepare them both for enduring and emerging
policy challenges.
As Ambassador Fick noted, our technology leadership is
stronger when we set a compelling example. With that, I would
like to share four important insights from our journey thus
far.
First, at State we see a booming demand for data and AI
services across the Department. Over the past 3 years, the
Center for Analytics has received over 350 project requests
from all corners of the Department. Some of them promote
foreign policy objectives while others bring about operational
efficiencies.
Second, on the talent front we are positioning federal data
science skills as close to the mission as possible. Our efforts
range from hiring data scientists to offering extensive AI-
related training for Department employees at all levels.
Third, we are already elevating our diplomacy and enhancing
operational efficiencies. We are not starting from scratch. For
example, by including a data scientist at COP-27 last year, the
U.S. delegation had access to real-time modeling on other
countries' proposed policy changes, which elevated our
country's negotiation position.
As another example, one of our partner bureaus, the Bureau
of Complex Stabilization Operations, employs computer vision to
document war crimes in Ukraine using satellite imagery. In
another area, the Bureau of International Organizational
Affairs employs AI to better strategize our approach within the
United Nations both in elections and votes resolutions.
These examples show how data and AI augment our diplomatic
capabilities. We also use AI as a force multiplier to increase
operational efficiencies. An example is the statutorily
required declassification process.
Traditionally, a resource intensive manual task, the
declassification process is now accelerated by machine
learning, achieving over 97 percent accuracy and reducing human
effort by 60 percent.
Last but not least, our AI advancements must rest on a firm
foundation of sound and responsible policy and governance. To
achieve this we formulated the Department's first AI policy,
updated our AI use case inventory, and launched an AI toolkit.
We have also appointed a responsible AI official and
initiated an AI steering committee to focus on upholding AI
principles and federal guidance.
As AI continues to evolve, our commitment to responsible AI
use remains unwavering. As we promote a vision of responsible
AI globally, the Department must and will lead by example.
Secretary Blinken believes that our workforce is more equipped
to lead globally when provided with timely, relevant data.
AI enhances this data's power, unlocking our workforce's
utmost potential. We are committed to harnessing AI's potential
and we are committed to doing it in a manner that is safe,
secure, and trustworthy.
Thank you for your attention. I am eager to address your
questions and collaborate on the promising future of
responsible AI at the State Department.
[The prepared statement of Dr. Graviss follows:]
Prepared Statement of Dr. Matthew Graviss
Chairman Cardin, Ranking Member Risch, and distinguished Members of
the Committee, thank you for the opportunity to speak with you today
about the role of artificial intelligence (AI) in modernizing our
diplomatic work around the world. I also want to recognize and thank
the Committee for its interest and the work it is doing on this very
important topic.
As the Department's first Chief Data & AI Officer, I lead the
Office of Management Strategy & Solutions' Center for Analytics (CfA),
which is spearheading efforts to use and scale responsible AI across
the Department. The Center for Analytics is the Department's hub for
data and applied AI, and we are leading in the implementation of the
Foundations for Evidence-Based Policymaking Act, the new Executive
Order on Safe, Secure, and Trustworthy Artificial Intelligence, and the
Advancing American AI Act. Our stellar team of data scientists,
engineers, and policy analysts provide support to bureaus, offices, and
overseas posts on priority projects, while also working with partners
across the Department--including the Foreign Service Institute and the
Bureau of Global Talent Management--to increase our collective data and
AI capacity. As Ambassador Fick mentioned, the work of my team promotes
AI internally at the Department, while his work is largely externally
facing and focused on the Department's engagement with the
international community.
President Biden's recent Executive Order on Safe, Secure, and
Trustworthy AI makes clear that ``AI can help government deliver better
results for the American people,'' and the State Department is no
exception. Just last week, Secretary Blinken signed the State
Department's Strategy for AI-Powered Diplomacy, an important milestone
in support of the Department's Modernization Agenda. The AI Strategy is
an essential step to equipping the Department's world-class diplomatic
corps with the analytics, training, safe infrastructure, and effective
policies to execute efficiently on enduring and emerging policy
challenges. As my colleague Ambassador Fick pointed out, our leadership
in technology foreign policy is stronger when we lead by example. That
is why his office and my office work so closely together to share what
we are learning and exchange information on future developments in AI
technology at home and abroad.
It is an honor to be here today to share four key messages. First,
our work responds to the high demand for data and AI needs from across
all corners of the Department to meet foreign policy objectives.
Second, we upskill and empower our workforce at all levels to deliver
on behalf of the American people. Third, the Department has tested and
proven approaches to turn data and responsible AI into insights and
efficiency gains. Fourth, the Department is prioritizing the ethical
and responsible deployment of AI to both seize its promise and manage
its risks.
On the first, we witness every day how the demand for data and AI
across the Department continues to grow. We have delivered AI and
advanced analytics projects in collaboration with over 48 bureaus and
offices, with the majority of projects directly supporting U.S. foreign
policy objectives, and the rest focused on expanding the operational
efficiency of the Department. The demand continues to grow. In the last
3 years, we've received over 350 requests for support, demonstrating
the significant interest in modern data and AI solutions across the
Department's components.
Second, through our workforce development initiatives, we are
placing federal data science and AI expertise as close to the mission
as possible. My office has led two Department-wide hiring initiatives
for data scientists, hired bureau-level chief data officers, and
created a standardized locally employed staff data scientist position
to enable embassies and consulates worldwide to hire expertise easily
and efficiently. Additionally, with a focus on broadening data analysis
and AI skills, Department employees have taken 62,000 hours of related
training, and our team has incorporated data modules into foreign
service tradecraft courses and Chief of Mission onboarding seminars.
This success demonstrates the tremendous value of President Biden's
National AI Talent Surge for increasing government AI capacity.
Third, Data and AI have already impacted American diplomacy by
equipping the right people with the right insights at the right time to
use information to provide advantages in international negotiations.
The Special Presidential Envoy for Climate did exactly that last year
at COP27, where the United States included a data scientist in its
delegation. This decision allowed us to model the impact of other
countries' proposed policy changes on environmental conditions in near-
real time while negotiations were ongoing--helping us hold countries to
account and push for ambitious commitments to achieve climate goals.
Elsewhere, the Bureau of Conflict and Stabilization Operations uses
computer vision and machine learning to document war crimes in Ukraine
through commercial satellite imagery; and the Bureau of International
Organization Affairs analyzes votes in the United Nations to better
target our resolution and election priorities. In these and many other
examples, data analytics and AI are strengthening the capabilities of
our world-class diplomatic corps.
But we're doing more than providing a decision advantage to U.S.
foreign policy practitioners; AI and data analytics are also making the
Department more efficient: for decades, the statutorily required
declassification process has been entirely manual, requiring thousands
of staff-hours and millions of dollars to execute. We and our partners
in the Bureau of Administration developed a machine learning platform
to accelerate this process by predictively marking cables for
declassification, continued classification, or further human review. In
our testing, the model has been over 97 percent accurate and over 60
percent faster than manual review alone--even accounting for the time
needed for human oversight. As we continue to put this model to use, we
are seeing the error rate improve even more. In addition to saving
thousands of staff-hours per year, this approach also safeguards
national security information from erroneous public disclosure. With
this solution, we are able to review every cable by statutory deadline,
as by statute, documents that are not reviewed on time are
automatically declassified.
Fourth, none of these burden reducing or foreign policy gains would
be possible without good data quality and responsible and trustworthy
AI policy and guidelines. In the past year, the Department has
published the first AI policy, updated its AI use case inventory, and
established a Responsible AI toolkit. We have also appointed a
Responsible AI Official and launched an AI Steering Committee, the
Department's governance body charged with advancing responsible AI
principles and ensuring adherence to federal AI guidance. As we
continue to take advantage of opportunities to leverage the power of AI
and the technology evolves, we will actively learn from our experiences
and that of our partners and outside experts to refine our policies and
ensure we have appropriate protections in place for its use in the
Department.
As the United States promotes a vision of responsible AI around the
world, we believe the Department must lead by example in leveraging AI
responsibly in our own work. As Secretary Blinken has said, our
workforce is ``better prepared to engage diplomatically, manage
effectively, and lead globally'' when it has the data it needs when it
needs it. AI and advanced analytics enhance the power of data and
accelerate our workforce's potential. We are absolutely committed to
pursuing U.S. foreign policy and operational advantages at scale in a
safe, secure, and trustworthy way.
Thank you very much for inviting me here today. I look forward to
answering your questions and to working with you to realize the vast
potential of responsible AI.
The Chairman. Well, thank you again, both of you, for your
testimony. We will go through 5-minute rounds.
Mr. Ambassador, you pointed out that America is the leader
in innovation and technology and we accept that, but I do not
want to be naive about it, and Dr. Graviss, you are talking
about how you are implementing it in our own use in our
Department.
We have seen this before. We have led in technology
development only to see the autocratic regimes that do not play
by the same rules that we do, try to steal our information, and
although we will proceed with--both on cyber and on AI within
our values, our adversaries do not share that commitment to our
values.
There are lots of tools in our toolkit--executive orders
that you have already mentioned and the executive orders and
treaties. I am just interested as to how you see these
standards being developed.
You talked about voluntary standards by what we do. That
works up to a point, but we know that the PRC is not going to
be as nice as we are in the protocols and adhering to
standards.
Where do we go as far as establishing international
guardrails? Are we looking at just America leading through
example or are we looking at more enforceable ways to establish
international guardrails?
Mr. Fick. Thank you, Senator.
Let me share a perspective on that and try to highlight a
handful of things.
First, we started with the voluntary commitments for two
reasons. First, voluntary by definition on the part of the
companies does not constrain innovation and, again, we are in a
race with our geopolitical adversaries and we cannot afford to
hamstring our innovative competitive advantage.
Voluntary equals innovation. Voluntary also equals speed.
The voluntary commitments are not intended to be the last step
in our domestic or our international governance structure.
They are a first step and they do a couple of things. They
allow us to build international consensus around a fairly
robust set of commitments on AI safety, security, and trust.
Safety means ensuring that the models do not return the most
dangerous results.
Security means ensuring the cybersecurity and other
integrity of the models themselves, and trust means helping to
ensure that users, the consumer, can distinguish between AI-
generated content and not AI-generated content.
The Chairman. That is fine for our allies that we have a
relationship with, but the autocratic competitors we have,
whether it be the PRC, whether it be Russia, whether it be
North Korea, they are not following those guidelines.
What protections are you negotiating that will provide
either incentives or disincentives for those who want to
violate international protocols?
Mr. Fick. Yes. I will give you three things.
First, when you are running a race sometimes it is
important to simply run faster than your competitor so that is
our innovation prioritization. Speed is one answer to your
question.
A second is trying to constrain adversaries who have a
wildly different view of the role of technology in our future
via mechanisms like export controls on our most advanced
semiconductors.
That is a small yard high fence approach in order to help
keep the most capable enabling technology of AI out of the
hands of our adversaries.
Then third, is a robust global engagement to shape the
norms of responsible state behavior in these critical
technology areas so that our adversaries can either abide by
those norms or put themselves and their behavior outside the
norms.
We are under no naive illusion that our adversaries are
going to comply with our norms, but building a broad coalition,
setting the normative example at least puts our adversaries
outside that framework so then we have the legitimacy and the
moral authority to call them out on it.
The Chairman. I would just point out we have gone through a
series of congressional actions encouraging administrations,
sometimes requiring administrations to take affirmative action
against bad players internationally.
We call them sanctions. We call them other issues. That is
pretty direct when we are dealing with specific actions that
are being taken--kinetic actions.
It may not be as easy to determine with the use of AI
tools, but if an adversary is not identifying the source or
using it for disinformation that undermines America's national
security, we are going to have to have more direction on how we
can assist or we may have to try to do that on our own.
Any help that you can give us as to how America can not
only lead in the race--we want to be first in the race--but to
prevent adversaries from taking us in the wrong direction or
stealing our information.
With that, let me turn it over to Senator Risch.
Senator Risch. Thank you, Mr. Chairman.
I am going to follow along the same lines the chairman did,
and that is if you think back probably some decades ago, some
congressional committee was holding a hearing just like this
saying, what are we going to do about this new thing called the
internets, as George Bush called it. How are we going to police
it? How are we going to--because it has tremendous possibility,
but there is evilness that that can arise from the use of it,
et cetera.
Frankly, after listening to you I suspect probably we are
in no different position here and that is whenever you talk
about voluntary, anybody here think that China or Russia are
going to voluntarily comply with any international norms when
they think it is in their best interest to do otherwise?
You got a country like Russia that cannot even obey the
simplest mandate of the United Nations; not to abuse one of
your neighbors.
China, look at them as to what they do with international
norms--international norms or as we call them, on technology
and patent protection of those kinds of things.
It is just stunning that they have no boundaries. They
are--it is just the Wild West out there.
I guess, Ambassador and Dr. Graviss, if you would comment
for a minute on are we right that we are probably barking up
the wrong tree? We talk in these esoteric terms about
voluntariness and goodness and kindness. Are we in the same
place with this as we are going to--as we have found ourself
with the internet?
Why do you not start, Ambassador?
Mr. Fick. Thanks, Senator. Just one point of clarification.
The voluntary commitments are voluntary on the part of the
companies. These are the leading AI developers in the world
subscribing to these commitments and committing their companies
to the responsible development of AI.
Senator Risch. I am assuming that you are going to want
that to flow over to----
Mr. Fick. We want international--We want businesses around
the world to support the commitments and then we want the
commitments to become the basis of codes and frameworks for
governments, and we are--please understand we are not naive
about whether our adversaries will voluntarily comply.
I think we need to confront an uncomfortable reality in the
software era, which is that controlling access to these
technologies is somewhere between very difficult and
impossible.
If an isolated and impoverished North Korea under strong
sanction could develop nuclear weapons, I would suggest that
the development of sophisticated software capability is a lot
easier than that.
Senator Risch. Good point.
Mr. Fick. Unfortunately, I do come back to our greatest
source of strength being maintaining our innovative advantage,
running faster, stewarding our competitive innovation
ecosystem, putting export controls and other sanctions in place
on adversaries where we think it can be effective and trying to
shape the global normative framework governing these
technologies.
Senator Risch. I appreciate that.
Mr. Fick. Thank you.
Senator Risch. Dr. Graviss, maybe you could take a crack at
what the AI police are going to look like.
Dr. Graviss. I can speak to what we are doing within the
State Department. That is my remit is focusing on internal, so
I will stick with that.
Senator Risch. Fair disclaimer.
Dr. Graviss. Just to point out we have the ability to set
policy and we have the ability to set policy quickly within the
Department and that is the first element.
We have already established an AI policy. We brought on
board a responsible AI official whose job is to be laser-
focused on the ethical and responsible use of AI. She has
brought on a test and evaluation team--a Red Teamer. She has
got these kind of capabilities on her team to provide
independent assessments, independent evaluations of AI that we
onboard in the Department. We have that policy lens and then we
also have that independent look that we are focused on and we
balance that with the opportunities.
There is a lot of repetitive work that is done all over the
world by our diplomats. We have to move swiftly and we have to
move smartly.
Senator Risch. Thank you. I guess it is going to be
interesting to see how this develops because I suspect that
when the internet started, probably people were a lot more
optimistic that that there was going to be compliance and you
would not have sovereigns actually and intentionally
interfering with other people's elections as they do today. It
will be interesting to see how this develops.
Unfortunately, like I said, where you cannot get countries
to behave themselves in public, how you are going to get them
to behave themselves in a laboratory with the door closed and
locked and nobody looking?
It will be interesting to see how that develops and I have
no doubt that you will follow it closely.
Thank you, Mr. Chairman.
The Chairman. Thank you, Senator Risch.
Let me just caution--we will be patient to wait for your
recommendations as to how we can protect those that are not
following guardrails, but the next person I am going to call
on, if it was not for his action in regards to Iran, we may
still be waiting for some executive actions.
Senator Menendez.
Senator Menendez. Thank you, Mr. Chairman.
President Biden and Chinese President Xi Jinping are set to
meet today during the APEC summit to discuss a number of issues
including the role of artificial intelligence in nuclear
weapons.
When I think about this topic, I worry that some country
would give the decision as to whether or not to launch a
nuclear weapon to artificial intelligence, thinking that it may
be the most astute way to make a calculus.
For me, that would be alarming. There have been press
reports that the United States and the PRC will pledge to limit
the use of AI in nuclear weapons systems.
Ambassador Fick, can you share with us why an agreement on
these issues with a country like China is significant? Is this
a model that the Administration plans to expand to other
willing countries?
Mr. Fick. Thank you, Senator.
I am serving at the State Department in part because of my
formative experience in the Marines in my twenties, which left
me with a conviction that diplomacy is and must always be the
nation's tool of first resort and so I welcome the conversation
between Presidents Biden and Xi.
I welcome communication between us and all countries at all
times. That does not mean we agree. It does not equal
collaboration, but the communication in and of itself is
valuable and is better than the alternative.
With respect to the military uses of AI, I would point to
the political declaration for the responsible use of AI, a
military focused framework of principles that rights-respecting
countries have subscribed to.
It is a good example of the kinds of behaviors that we want
to become the global norm. Again, we are under no illusion that
our adversaries will abide by those norms in every case, but I
think that the channel of communication in itself is valuable.
Senator Menendez. I have no doubt about the communications.
I agree with you totally. I have been advocating for that so
that we have lines of communication, particularly for
nonconflict issues in China.
My question is really going to, do we want to extend an
understanding and agreement as to how or whether or not even AI
should be used at all particularly in terms of nuclear
launches, as well as how do we ensure--because China is growing
their nuclear capabilities dramatically--how do we ensure that
even if a country signs on to that how is it--have we thought
about the enforcement mechanisms of that?
Mr. Fick. I think one of the foundational principles of the
political declaration is a belief that when we are talking
about lethal systems of any kind, there needs to be a human in
the loop.
We cannot have fully machine-enabled or AI-enabled lethal
systems. Your question about the inspection mechanism and the
enforcement mechanism is, of course, exactly the right one.
I would have to refer to colleagues elsewhere in the
Department who have that piece within their remit.
Senator Menendez. All right. China has introduced some of
the earliest and most detailed regulations governing artificial
intelligence over the last 2 years. Our own allies, including
the European Union, have also begun cooperating on developing
regimes to govern the proliferation of AI.
While I applaud these engagements on global AI governance,
more must be done in this area of strategic competition if we
are to keep up with China's efforts to regulate AI and provide
an alternate vision for AI that embraces human rights, privacy,
and other values we hold so dear.
What are some of the troubling implications if America and
our allies fail to develop a consensus around regulating AI and
we allow China to pioneer the rules of AI globally?
Mr. Fick. I would make two points there in response to your
question, Senator.
First, our foreign policy in any area will rarely be any
stronger than our domestic policy in that area. American
foreign policy--I think the foreign policy in any transparent
democratic society is a reflection of what we do at home, and
so we do need to get our own house in order with respect to the
regulation and governance of these emerging technologies so
that we are presenting on the world stage a responsible
framework for governance that has legitimacy and moral
authority.
Second, there is no room here for one of the concepts that
even our closest European allies advocate, which is digital
sovereignty. I would argue we need to think about digital
solidarity. We have to think about close collaboration with our
closest allies and partners to coordinate R&D investment, to
coordinate regulatory interoperability.
Companies want large harmonized markets. We need the
combined integrated efforts of our best universities, of our
students and workers. It points to the importance of our
domestic regulatory regime. It also points to the importance of
our allies and partnerships in order to both shape the
international normative environment and also present the
strongest unified front against the PRC model.
Senator Menendez. Mr. Chairman, if I may.
I am going to just submit a question for the record since
my time has expired, but I would like really a responsive
answer to it. This is the 10th anniversary of the Rana Plaza
factory collapse in Bangladesh. Took the lives of over 1,000
workers.
In this AI industry, workers are being exploited, paid as
little as $1 an hour, going through the repetitive process of
trying to determine what is a smile, what is a frown, to inform
the algorithms, and I would like to see us lead in this regard
to make sure that we do not have worker exploitation globally
in the development of AI. I would like to see what the
Department is doing with that.
Thank you, Mr. Chairman.
The Chairman. I would be interested in that answer also. As
I mentioned, we need to lead by our values.
Senator Ricketts.
Senator Ricketts. Thank you, Mr. Chairman, and thank you to
our witnesses here today.
Ambassador Fick, I want to build on what Senator Menendez
was talking about with regard to the EU specifically.
Obviously, we have talked about the different sorts of
regimes--regulatory regimes that different countries around the
world are trying to establish. We talked about the PRC. We have
talked about what we have done with regard to the Executive
Order and the voluntary guidelines you talked about.
Then, of course, we got the European Union where they have
been working on their own legislation. They proposed the AI Act
a couple of years ago and it involves a risk-based approach.
As you know, the European Union will often come up with
these standards and then companies will really adopt those for
their worldwide products to try and standardize their products
and make it easier to have access both to the market in Europe
and around the world.
If you have--I understand the legislation is coming to its
final phases. They have not figured out the generative AI part
of it. Assuming that gets passed, companies will be forced to
make a decision of either complying with what the EU does or
creating completely different algorithms that do not have
access to EU data and then having to deal with the difficulties
of that as well.
My question really gets to if we have got these EU
regulatory constraints--you talked about cooperating with our
allies, but they are moving forward on this legislation. I
think you have been critical of this legislation. What effect
if the EU passes their AI Act will that have on the United
States and our other allies and is this going to potentially
put a damper on that innovation you have talked about that is
so important among U.S. companies?
Mr. Fick. Thank you, Senator. Yes, we have been in regular
dialogue with our EU counterparts, both member states and the
European Commission, on the substance of the AI Act.
I would make three points on this. The first is speed. As
you pointed out, the Act was first drafted before ChatGPT was
released a year ago and so in some sense it has already been
overtaken by events, which points to----
Senator Ricketts. It is already obsolete, essentially.
Mr. Fick. --a structural challenge with that kind of
regulatory approach. It also as currently drafted would hinder
law enforcement cooperation between the United States and the
EU, things like the use of facial recognition technology at
borders or by law enforcement.
Third, we hear repeatedly from companies and innovators not
only in the United States, but also in Europe that is not
adequately protective of intellectual property.
It would deter entrepreneurs from building AI businesses in
Europe. Back to this point on digital solidarity, the future
that I think we think positions us most strongly for
competition with our adversaries, is one where we have globe-
spanning technology businesses being built not only in the
United States, but also in our closest and rights-respecting
allies and partners in Europe and in Asia.
Senator Ricketts. For all those reasons you just talked
about, how we want to be working with our allies in Europe, and
yet they are pushing forward with this, frankly, it is already
obsolete now or at least it is not complete legislation, what
is the State Department doing to get them to maybe to slow this
down, to be able to work with us more carefully, to create that
regime you are talking about where we can be on the same team
with regard to pushing back against the People's Republic of
China and not creating an obstacle to us working together to
oppose them?
Mr. Fick. We are pushing in every form on this,
multilaterally and bilaterally. I will give you three examples.
Over the last 3 weeks, I met with the German communications
minister who has oversight of these policies in Germany when he
was here in Washington. He was at Bletchley Park. Most of our
EU counterparts were there and last week I was in France with
President Macron himself and the French digital minister making
exactly these cases, that the continental Europe in a lot of
ways regulated itself out of the cloud services era of
technology.
It is no accident that the five largest global cloud
services businesses are here in the United States. It would be
to our mutual detriment if the EU were to regulate its way out
of the AI era of technology innovation. We are pushing on this
in every form.
Senator Ricketts. Okay. Then you just said you were at
Bletchley Park. How are the Brits seeing this? What is the U.K.
policy going to be and where are they going? Are they more
leaning toward the European model or are they leaning more
toward our model?
Mr. Fick. Prime Minister Sunak and his government are
leaning very strongly toward our model. I think we and our
British allies are very close on this. They intend to set
themselves up as an AI superpower, in their words, which we
view as very much to our benefit as well.
Senator Ricketts. Great. Thank you very much. Thank you,
Ambassador.
Mr. Chairman, thanks.
The Chairman. Thank you.
Senator Shaheen.
Senator Shaheen. Thank you both for being here.
Ambassador Fick, in your opening statement--and you
mentioned an alphabet soup of multilateral organizations that
we were looking at--you did not mention NATO, and while I
recognize that NATO is not technically part of the State
Department or within the purview of the State Department,
certainly what they do is very important.
Was there a reason and what are you looking at with respect
to NATO?
Mr. Fick. Senator Shaheen, I joke with my wife sometimes
that we should get an apartment in Brussels because I am at
NATO so much, especially in the context of the war in Ukraine
in the past year.
Our ambassador to NATO, Juliannee Smith, is a dear friend
and we collaborate very closely on these topics. I think that
NATO is a terrific example of what Secretary Blinken talks
about as variable geometry and making sure that our most
important alliances historically are fit for purpose in the
technology age.
A really key element of our diplomacy with NATO and in
Europe has been ensuring that the alliance is equipped on all
of these technology topics. Cybersecurity, at the Vilnius
summit in July, we saw a robust set of deliverables on the
virtual cyber incident support capability, the revitalization
of the NATO cyber defense pledge.
We have been working with allies across the alliance to
ensure that their own architecture and infrastructure is secure
and trusted so that we can share information and intelligence
freely across the alliance.
That was not an error of commission in not mentioning NATO
in my opening comments. I think in technology, the NATO
alliance is one of our leading partners globally on all of
these topics. Thank you.
Senator Shaheen. Good. Thank you.
As everybody has talked about so far, AI has tremendous
potential in so many areas, but it can also perpetuate existing
biases and inequities. How does the Department look at how we
ensure that those inequities are not part of whatever is done
through AI, particularly with respect to gender or with respect
to the whole range of ethnic racial differences that we want to
respect?
Dr. Graviss. Thank you, Senator, for the question. I will
make a few points here.
One is it starts with training, leveling up the literacy
when it comes to data, when it comes to AI, when it comes to
algorithms across the Department. We understand that this is a
possibility.
The second is using good data. Good AI rides on good data,
at the end of the day, and so the focus of ours is to how do we
bring this kind of technology into our Department in a secure
infrastructure and apply this technology on data that we
manage, data that we can ensure the quality of. That is the
second point.
The third has to do with our partnership with industry,
which is really focused on the procurement process--how do we
ensure that software procurement includes the right policies
and requirements to make sure that our industry partners are
providing technologies that are free of those types of biases
you mentioned.
Senator Shaheen. Great. I want to be a little more
parochial now. In my office, the number-one constituent issue
we hear from people about are visas, passports, challenges with
our immigration system. Yes. The State Department particularly
during COVID had some real issues in countries around the world
in terms of processing those kinds of issues.
Tell me how AI is going to help us make sure we can process
those kinds of constituent concerns faster so that we can
provide service to people who need it.
Dr. Graviss. Passport--the passport organization within
Consular Affairs is not using artificial intelligence at this
time.
Senator Shaheen. I know they are in New Hampshire or at
least one of them is. We are very aware of that.
Dr. Graviss. We will make that point first.
More broadly, we see this as a real opportunity to reduce
the type of repetitive rote work that happens across the
Department, both domestically and overseas. We are taking a
very pilot-centric approach, starting small, learning a lot,
scaling where it works and we have seen some real success with
that approach.
Senator Shaheen. Assuming that you continue to see success,
what do you need in order to ramp that up? Do you need
resources? Do you need more people? Do you just need the IP to
share those with our embassies around the world?
Dr. Graviss. I am a systems engineer, so what goes in can
increase the output. In my mind it is resources in terms of
infrastructure and in terms of personnel.
It is the people without the actual technology, without the
data to be able to employ it are not going to get very far in
technology without the skilled workforce that we are currently
aggressively trying to bring into the federal employment within
the State Department. You got to have both.
Senator Shaheen. Thank you. Thank you, Mr. Chairman.
The Chairman. I assure you Senator Shaheen has the support
of every member of this committee on consular services to be
streamlined and made more timely. It would make our life a lot
easier. I assure you of that.
Senator Barrasso.
Senator Barrasso. Thanks, Mr. Chairman.
For either you who want to address this, I have with me
today the principal and a number of teachers who are here from
the Southside Elementary School, Powell, Wyoming. They are a
Blue Ribbon school, one of only two in the state of Wyoming,
and they are here for the national celebration of their
achievements and this has to do with AI, TikTok. Whoever you
want to--whoever wants to answer it, because reports do
indicate this--the TikTok is harvesting vast amounts of data
from millions of users in the United States, many of whom are
students.
It is owned by a company based in Beijing. Although TikTok
executives deny it, many experts believe that the Chinese
Government has access to TikTok's American user data.
The question is do you believe that TikTok is using data
collected from its application to feed Chinese artificial
intelligence, the machine learning algorithms and what the
implications are for us here at home and for students all
around the country?
Mr. Fick. Thanks, Senator. A couple of couple of points on
that.
I would point out, of course, that the Administration has
enacted a ban on the installation of TikTok on federal devices,
which should speak to, I think, our perspective on the risk
involved.
The second observation, I am probably the least popular
father in my daughter's middle school because of my personal
views on this point and my request to you would be that we try
to speak with one voice as a country, that we adopt a position
via a democratic process that acknowledges what you have said.
Senator Barrasso. Anything you would like to add on that?
Dr. Graviss. Thank you, Senator. I would say we are worried
about our data, too, at the State Department. How do we ensure
that that our employees are not putting sensitive information
out on the web?
We have existing policies in place that need to be
reinforced and training and literacy has to go along with that,
and then we need to move aggressively to onboard this kind of
technology into the infrastructure that we micromanage, that we
have tight security controls on and that we are constantly
monitoring.
Senator Barrasso. Thank you.
Mr. Fick, China continues to infiltrate top U.S. companies,
laboratories, universities to I believe steal valuable American
intellectual property. The impact on the economy is massive,
the Chinese Communist Party clearly attempting to surpass
America in terms of our economic strength, military strength,
international influence.
A senior FBI official had said earlier this year, nation-
state adversaries, particularly China, pose a significant
threat to American companies and national security by stealing
our AI technology and data to advance their own AI programs and
enable foreign influence campaigns.
How do you assess the threat of China's increasingly
adversarial government to the United States?
Mr. Fick. Senator, I certainly agree with that assessment,
that intellectual property theft by the PRC has stripped
critical, competitive advantage out of the U.S. economy for
several decades now.
Our response to that or uneven response to that has put us
in a deterrent hole and there is a very real threat to the
security of our most advanced AI systems, going forward.
It is why one of the pillars in the voluntary commitments
is the security pillar, which is standards around
cybersecurity, testing, disclosure of testing results for our
most advanced frontier models and a very real shared focus by
the government and these companies on safeguarding that aspect
of our innovative advantage.
Senator Barrasso. Then a final question, since Senator
Risch started by talking about the leadership that we have
taken in the United States when it came to the internet years
and years ago, we led in this sector because we chose to take a
kind of a light-handed regulatory approach.
A number of us in the Senate have met with--we have had
discussions on AI with Elon Musk, Mark Zuckerberg, a number of
the leaders in the technology there.
Do you believe that a light-handed regulatory structure or
a heavy-handed structure, if you will, is best for promoting
American leadership and innovation in artificial intelligence
and machine learning technology?
Mr. Fick. Senator, I have been in this job for a year, but
I spent a dozen years before that building a software business
and investing in software businesses, and I believe very
strongly in the power of our innovation economy as our greatest
source of national strength.
I said in my opening statement that many traditional
measures of strength like GDP and military capacity are
increasingly downstream of our ability to innovate in these
technologies.
I do think there is a role for appropriate regulation of
such powerful tech and I think our North Star must be
maintaining the health of that innovation economy.
Senator Barrasso. Thank you, Mr. Chairman.
The Chairman. Senator Barrasso, thank you for bringing the
students here today. I think the students should know that
Senator Barrasso is one of our leaders in the United States
Senate, but his role here on the Senate Foreign Relations
Committee is invaluable. He is a great member of our committee.
As Dr. Graviss has said, there is a career at the State
Department in AI, so applications will be given out shortly.
Thank you for being here. We appreciate it very much.
Senator Van Hollen.
Senator Van Hollen. Thank you, Mr. Chairman. Thank both of
you for your testimony and for your service, and I know we have
covered a lot of ground today, including some of the questions
that I was going to ask.
I just want to affirm a couple of principles, Ambassador
Fick, that I know you have outlined not just this morning, but
earlier, which is in order--that part of our overall digital
and data strategy has to be to promote American innovation,
support American values including promoting human rights. Would
you agree?
Mr. Fick. Yes.
Senator Van Hollen. You would agree, I think, that
authoritarian regimes fear the open flow of information and
data and try to put up walls and barriers against it, right?
Mr. Fick. I agree.
Senator Van Hollen. Would you agree that one of the
mechanisms they use to do it is to try to promote data
localization requirements around the world so that those
governments can sometimes spy on their people and look for
dissenters?
Mr. Fick. We see an increasing balkanization of the free
and open internet.
Senator Van Hollen. Here is my question, because we have a
unified executive branch. Were you consulted when the USTR
decided to back away from the e-commerce principles at the WTO?
Were you consulted as part of that decision?
Mr. Fick. I read about the decision in the press, Senator.
Senator Van Hollen. I know what your earlier testimony--one
of the things that struck me when you were up for your
nomination was that you were a straight shooter.
Would you agree that that decision undermines the
principles you and I just discussed about an open internet and
not allowing authoritarian regimes to more easily spy and
crackdown on dissent?
Mr. Fick. We advocate in every forum for the benefits of an
internet that is free and open and interoperable and reliable
and secure and that includes the free flow of data.
Senator Van Hollen. I appreciate that, and I understand
that you are here not just as a State Department
representative, but part of an administration.
I appreciate that answer because I can tell you that, in my
view, that what the USTR did at the WTO was--totally undermined
the principles that we just talked about--free flow of
information, stepping back from our resistance to data
localization, which empowers authoritarian regimes, as well as
our efforts to prevent forced tech transfer.
Those were all principles that we advocated for and it is
my belief that when we back away from that, we lose our
credibility around the world, not just on these issues, but on
the issues that you two gentlemen have to work on every day.
Mr. Chairman, I would just like to submit for the record a
letter that a group of us wrote to President Biden on this
issue.
The Chairman. Without objection.
[Editor's note.--The information referred to above can be found
in the ``Additional Material Submitted for the Record'' section
at the end of this hearing.]
Senator Van Hollen. Thank you.
On the issue of AI, of course, one of the big issues is,
and I think it has been discussed a little bit already this
morning, the issue of international standard-setting
committees, and there are many, but there are a number of key
ones.
How would you evaluate the current U.S. presence,
participation, influence in those committees, and I would start
with Ambassador Fick and then hear from Dr. Graviss.
Mr. Fick. May I just ask for clarification on which
committees, Senator? Which organizations?
Senator Van Hollen. Yes, there is the international
standard-setting committee being one of them. There is one on
electrical standards. Those are----
Mr. Fick. Okay. Yes, I understand.
Standards harmonization is one of the foundational pieces
of building an innovation ecosystem across national boundaries
that can bring to bear shared R&D, shared innovation, more than
the companies of any one business. We are strongly engaged in
those international standard-setting bodies to help ensure that
the standards are aligned with our values, that promote shared
prosperity, and that they--looking ahead now to the AI era that
we adopt a similarly legitimate, broadly consensus-based
approach to harmonizing standards.
Senator Van Hollen. I appreciate that. I think the major
ones for the purposes of AI, I think, are ISO, the IEC and the
ITU, but there may be others and I think--as you know, the PRC
and others have been very actively engaged in trying to
increase their presence and participation. It has been one of
America's strengths that we have led that effort and I know you
both are committed to doing that.
I do not know, Dr. Graviss, if you had anything to add to
that question.
Dr. Graviss. Sure. Happy to. A couple of things, Senator.
First is within the Federal Government, starts with the
National Institute of Standards and Technologies. That is the--
who is setting the guardrails around what standards are
expected to be followed across the Federal Government. Also on
the cybersecurity side, Cyber Infrastructure Security Agency
within DHS.
At the State Department, as I just said, AI rides on good
data. We have been doing data standards for the last several
years. It is not the flashiest part of the work, but it is
necessary and I think it is great that you are talking about
that.
Also, I will point out that the International
Organizations--Organizational Affairs Bureau--IO Bureau within
the State Department has their own chief data officer. It is
part of a program that we have been leading to place data
leaders in each bureau and she is focused on analytics related
to elections try to get Americans in the right committees who
are setting these kind of standards and that is really
important, too.
That is a great example of how we apply technology, how do
we apply AI in the Department to impact on a global stage
diplomatically.
Senator Van Hollen. Thank you. Thank you, Mr. Chair.
The Chairman. I just want to comment briefly in introducing
Senator Kaine--make it clear, Mr. Ambassador, we strongly
support diplomacy in dealing with these issues. Our questions
are really aimed to seeing how we can make sure you have the
resources, the tools, and then sometimes the direction you need
for diplomacy to be successful.
We are very fortunate on this committee to have our own
ambassador to the Department of Defense. Let me explain that.
Senator Kaine serves on this committee and the Armed Services
Committee, so we have our representative with DoD to promote
diplomacy.
Senator Kaine.
Senator Kaine. Mr. Chairman, you anticipated my line of
questioning because what I want to talk about is something that
is at the intersection of the Foreign Relations Committee
jurisdiction and the Armed Services Committee jurisdiction and
that is the AUKUS partnership.
President Biden announced the AUKUS partnership with the--
Australia and the U.K. a couple years back. It has been fleshed
out. Last summer, the Administration came to us to ask for
enabling legislation to move the partnership forward.
They thought it was going to be in the Armed Services
Committee, but when I told them, no, all this comes through the
Foreign Relations Committee, they said, well, good--you are on
both, can you help out? Senator Cardin has been key to this as
well.
The partnership has two pillars. The first pillar is the
construction over a period of about 25 years of nuclear subs
that would be Australian-operated and eventually Australian-
built to provide additional stability in the Indo-Pacific.
Just for my colleagues, I was at a Navy-funded workforce
training school in Danville, Virginia, on Monday that the Navy
has stood up to build out the submarine industrial base.
Their 60 students go through in five different curriculum
areas for an 8-week program and students and--multiple students
in each of the five programs were from Australia.
They were not only from all over the United States, but
Aussies are already training to be part of the sub industrial
base and Aussie sailors are now part of the Nuclear Power
School that the Navy operates in South Carolina.
Even though the first subs are not going to be purchased
until the 2030s or built in Australia until the 2040s, we are
integrating.
The second piece of AUKUS is pillar two, which is
innovative technologies, quantum undersea capabilities, but
also AI, and I would like to ask your own impressions about
what we can do together with Australia and the U.K. in the
pillar two space to really maximize the synergies of us working
together on AI capacity that can help us in the defense of the
INDOPACOM.
Mr. Fick. Thank you, Senator. Happy to address that.
I think that my overarching view here is maybe reflected in
the way Senator Blinken talks about alliances and partnerships
that are fit for purpose, the variable geometry of our
alliances and our partnerships, and pulling on the right
partners in the right fora at the right time to accomplish our
highest priorities.
This is a good example. Specifically, in the innovative
technology pillar for AUKUS. I will point to one thing that we
announced just a few weeks ago during the Australian Prime
Minister's visit--I have been working closely with Ambassador
Kennedy on this for months. That is a partnership on undersea
cables connecting Pacific Island countries and what is
interesting here--I wish Senator Shaheen was still here because
one of the world's leading manufacturers of undersea cables is
in New Hampshire, and so we have a national champion in the
United States in this regard.
We have an opportunity now to lay trusted cable to connect
Australia and the United States, but also to link together
other key strategic geographies in the Pacific and do it in a
way that is part of a vertically integrated technology stack.
Let me just explain for a moment what I mean.
The cables as the baseline connectivity--those cables are
essential for the data backhaul if we were eventually to build
data centers, say, in Australia so that key countries across
the Pacific could migrate their government enterprise to the
cloud, could ensure continuity of communication and continuity
of operations in the event of a contingency scenario, and then
put top-notch, cutting-edge cybersecurity on top of those data
centers.
Again, a layered approach of trusted cables, trusted data
centers, and advanced cybersecurity. That is what we are in the
early stages of doing in partnership with Australia and others
in the region.
I think it is really exciting and it is a template that we
ought to be able to replicate now in other geographies.
Senator Kaine. My colleagues asked questions earlier that
you--that led you to talk about some of the different
approaches that the EU is taking, for example, compared to the
United States and the hope that the U.K. would be more aligned
with the U.S. in terms of AI regulation, less centralized
possibly.
My hope would be that as we are having those discussions
with the U.K., we also keep Australia in mind because I think
if we are going to make this pillar two part of AUKUS
successful, the greatest alignment between the U.S. and U.K.
and Australia on matters like AI regulation would be helpful.
I hope we might be able to convince the EU to come more of
our way as well, but if you have any thoughts about the
Australian approach to AI regulation, I would love to hear
them, but at least I would like to encourage that we include
Australia in the discussions that we are having with the U.K.
around alignment.
Mr. Fick. I agree completely, Senator. I was at Bletchley
Park for the U.K. AI safety summit a couple weeks ago. At the
end of the summit, we released the Bletchley communique.
Australia is a signatory there and so I think we are fully
aligned.
Senator Kaine. Great. Thank you, Mr. Chair.
The Chairman. Senator Young.
Senator Young. Well, thank you, Chairman, for holding this
very important hearing. I thank our witnesses for your service.
This is such an important topic we are all dealing with in
various ways.
I will dive right in. Ambassador Fick, I have a number of
questions for you. I ask that you keep your responses as tight
as possible so we can get through them.
As we think about harmonizing AI policies and R&D between
partners and allies, can you speak, Ambassador, to the
importance of forging clear and enforceable digital trade rules
that will ensure the regulated flow of data across
international boundaries that may enable future AI and machine
learning systems and which have to entail necessary cross
border privacy protections?
There is a possibility that a failure to reach these sorts
of rules and agreement on the rules may jeopardize AI
collaboration and cooperation.
Ambassador, do you believe it would be to America's
advantage to prioritize digital trade in our multilateral and
bilateral trade negotiations? Not a trick question.
Mr. Fick. Thanks, Senator. I was at Purdue a few weeks ago
in Indiana and let me let me just say that Keith Krach, who has
endowed a center for tech diplomacy at Purdue, in many ways was
my predecessor with his portfolio and put the markers down on
some essential topics and we have tried to carry that work
forward.
Yes. In a word, I think that harmonizing data flows and
prioritizing digital trade in our international trade agenda is
an important component of American competitiveness in the 21st
century.
Senator Young. How do we get there?
Mr. Fick. Through a series of bilateral/multilateral
negotiations that result in outcomes like the U.S.-EU data
privacy framework. There is no silver bullet here. I think it
is a lot of hard-fought diplomacy bilaterally and
multilaterally.
Senator Young. I agree with you, and I know others in the
Administration agree with that perspective as well because
trade is--at once it is a vehicle for upward mobility for
prosperity, but also for forging tighter relationships among
countries and regions and I think it is very important that
everyone in the Administration regard it as such.
We will continue to work on the Finance Committee and the
Foreign Relations Committee on the topic of digital trade.
Ambassador, can you speak to how your bureau is approaching
the use of artificial intelligence in converging technologies
like synthetic biology, especially in gene synthesis and gene
editing, as you promote the responsible adoption of artificial
intelligence?
Mr. Fick. I think the simplest answer to that question,
Senator, is that we have seen the PRC run what we now call the
Huawei playbook in telecom. It is a series of--has a series of
elements--IP theft, subsidies.
That playbook--that rough template is now being run by the
PRC across a whole host of other emerging technologies
including synthetic biology.
Senator Young. All right. Thank you.
There is much congressional discussion, Ambassador, around
the regulation of artificial intelligence. As the chairman was
kind to acknowledge earlier, I have been pretty involved in
many of those conversations and much of what we discussed has
centered on whether we can use existing statutes and
regulations for specific use cases of artificial intelligence.
I think this is probably the approach our country will take
on a going forward basis, but the same sort of debate about how
conceptually you approach AI regulation extends to the global
harmonization of policies across countries, across regions.
There have been countless forums already, and I think that
is a good thing, to discuss artificial intelligence from the G-
7 to the EU to a recent conclave on safety as it pertains to
frontier models that was in the U.K., and making sense of how
our country is going to approach harmonization can sometimes be
complicated and confusing.
From your vantage point, Ambassador, how is the Department
approaching this question of global harmonization? Is the
intention to pursue action through existing multilateral
institutions or instead to convene special multilateral
conventions or pursue bilateral agreements?
Then, in any harmonization effort, what role do you
envision the U.S. private sector actors are going to play?
Mr. Fick. Thanks, Senator. Yes, just in the last 10 days we
have seen four strands come together. We have seen the AI code
of conduct for developers at the G-7, the American executive
order, the Bletchley convening and communique, and the AI Act--
the EU AI Act in trilogue negotiation.
We have this dynamic policy landscape and our overarching
next step is to try to drive these to convergence in some sort
of a comprehensive policy framework that is going to be to the
advantage of rights-respecting allies and partners--the U.S.
the EU, the Japanese, the Australians, South Koreans--and as
inclusive as possible of middle ground states that are willing
to subscribe to a rights-respecting use of these technologies.
Senator Young. You have not settled--just by way of follow-
up, you have not settled on a particular format, you have not
identified who will sit in the individual seats around a table
as you try and harmonize different policies? Is that accurate?
I am not asking this critically----
Mr. Fick. Sure.
Senator Young. --but as an important investigative matter
by this committee.
Mr. Fick. I think there are two basic approaches we could
take. One approach would be to form a new group--a T-12, a T-
15, a T-18--an alliance of techno democracies, if you will.
That has been in the press, in the think tank world. It has
been a subject of discussion.
The other general approach is to use the alliances and the
partnerships that we have or to create new ones and make sure
they are fit for purpose--Senator Shaheen's question about
NATO, Senator Kaine's question about AUKUS.
My view and the view of the Department, the view of the
Secretary, is that we should do the latter--that we should be
using organizations that exist. We should not be setting up new
secretariat's with overhead, creating new sets of meetings for
people to attend.
We should be using existing organizations. We should be
ensuring that they are fit for purpose and we should be making
sure the technology issues and AI governance are threaded
through all of them.
Senator Young. Okay. I very much hope that you will keep
myself and other members of this committee informed of
developments as it pertains to who you will be convening, when
you will be convening them.
I hope it is not in the very distant future--that is, the
first quarter or two of next year--and what you anticipate the
agenda looking like.
Mr. Fick. Senator, if I may. You asked a question about the
private sector I think it is important to address here.
Senator Young. Yes, sir.
Mr. Fick. The private sector has been front and center in
these discussions from day one, starting with the voluntary
commitments of the White House. They have been involved in
every multi-stakeholder consultation in the G-7 and the G-20
and other fora. They were present in force at Bletchley Park,
widely consulted on the E.O.
Again, back to the point that the bulk of the innovative
horsepower in our economy sits in the private sector. The bulk
of our talent is in the private sector. The bulk of the attack
surface we care about protecting is in the private sector.
This only works if it is truly a multi-stakeholder process
involving governments, companies, and civil society and other
organizations.
Senator Young. Thank you, Ambassador. Thank you, Chairman,
for your patience.
The Chairman. Let me again acknowledge that Senator Young
has been instrumental in bringing together the top experts
basically in the world on AI in a nonpartisan environment where
we can really absorb and try to understand where we are today,
where we are heading, and what are our options and what are our
needs.
Thank you very much for that help. Appreciate it very much.
Senator Risch, anything?
Senator Risch. I am good. Thank you.
The Chairman. Let me thank our witnesses. Let me just
underscore the point. We recognize this is not just a
governmental exercise. It is the private sector as well, but
the private sector also needs to have guardrails and needs to
be protected.
That is why the global rules are incredibly important for
how the United States leads in establishing those global rules
not only for state actors, but also for private companies.
This conversation will continue. We anticipate there is
going to be legislative action throughout 2024 in regards to AI
to try to get a handle on appropriate guardrails and to give us
the tools so we can continue to lead in innovation, but also in
the responsible use of AI.
Thank you both for your contributions and we look forward
to the continuation.
The record will remain open through close of business on
Friday for members who may have questions for the record. We
would ask that you respond in a timely way.
Thank you.
Mr. Fick. Thanks for having us.
The Chairman. The hearing is adjourned.
[Whereupon, at 11:25 a.m., the hearing was adjourned.]
----------
Additional Material Submitted for the Record
Responses of Mr. Nathaniel Fick to Questions
Submitted by Senator Benjamin L. Cardin
Question. U.S. Leadership in Global AI Governance: The world has
recognized the need for international engagement and cooperation on AI,
underscored by the recent rapid advancements in powerful generative AI
models. The U.S. has played a leadership role in the global
conversation on AI, including at the AI Safety Summit in London and
through the G7 AI International Code of Conduct. As governments around
the world attempt to respond to the proliferation of these advanced AI
tools. What role should the United States play in shaping international
AI governance and what are our highest priorities? How can we ensure
that our interests and values are protected as other actors, including
some of our competitors, seek to shape the global AI landscape?
Answer. The United States is taking a strong leadership role in
shaping international AI governance. Our priorities include: leading
with an affirmative, positive vision on technology; maintaining a focus
on innovation; ensuring trust, transparency, and a risk-based
framework; establishing guardrails around AI use; mitigating potential
harms; and ensuring AI policy is grounded in established ethical
frameworks.
The United States promotes the responsible use of trustworthy AI in
line with democratic values and respect for human rights. We advocate
for an international, multistakeholder AI ecosystem that enables
innovation and fosters public confidence in AI. We can ensure that the
United States' interests and values are protected by building
coalitions of like-minded countries who coalesce around a set of
commitments on AI that reflect these values.
Question. How are U.S. domestic efforts around AI impacting U.S.
competitiveness and leadership internationally? Does the U.S. need
domestic regulation on AI in order to remain competitive on the
international stage?
Answer. The United States must continue our international
engagement on AI even while domestic efforts take shape. The Department
of State's role is to advance U.S. AI policy in the international
domain. This is an iterative process as the Administration works on its
lines of effort, such as the voluntary commitments and the Executive
Order, while discussions in Congress on possible legislation are
underway. President Biden has also called on Congress to pass
bipartisan data privacy legislation which will further support the
Department's AI-focused engagement internationally.
However, even as we and others work on domestic approaches, we
believe international dialogue is crucial to maintain interoperability
and build consensus around shared regulatory approaches. The U.S.
approach domestically to AI is directly relevant to our global
influence and how other governments choose to govern this technology.
Our multi-stakeholder approach begins with university and industry
research and innovation, includes voluntary commitments from leading AI
companies for safety and security, and incorporates civil society
stakeholder input. The United States seeks to be a global model of how
forward-looking, rights-respecting countries can leverage AI to advance
both our shared values and our shared aspirations by putting in place
the requisite guardrails to ensure safety and security and mitigate
harms.
Question. Does the Department require any additional resources or
authorities to promote U.S. leadership on AI?
Answer. With the creation of the State Department Bureau of
Cyberspace and Digital Policy (CDP) and the Office of the Special Envoy
for Critical and Emerging Technology, we are well placed to promote
U.S. leadership on AI. Due to the ever-increasing global interest in
policymaking on AI and related digital economic issues, and considering
the growing long-term geopolitical, economic, and human rights
implications, additional personnel and resources remain a necessity.
The establishment and funding of a dedicated Foreign Assistance Fund
for Cyber, Digital, and Related Technologies would help support U.S.
leadership in this space, represent a needed step forward in the
evolution of U.S. cyber diplomacy, and provide a clear return on
investment for U.S. national security interests. CDP remains eager to
discuss this important concept with Congress.
Question. AI and Strategic Competition: We are all very aware that
the PRC Government has already demonstrated its willingness to use AI-
enabled tools like facial recognition and social scoring to facilitate
mass surveillance and repression on its own population, particularly
Uyghurs and Tibetans. Given these facts, do you think it was
appropriate that the PRC participated in the UK AI Safety Summit? Why?
Can you help us understand the fact that the United States AND the PRC
both agreed to the same set of AI principles outlined in the Bletchley
Declaration at that summit?
Answer. The Bletchley Declaration represents an important expansion
of the multilateral work on AI we accomplished this year, including
through the G7 Hiroshima AI Process, with likeminded partners and
allies. The decision to invite the PRC was made by our UK counterparts,
who were the hosts of the conference. Many major stakeholders active in
the AI space, including countries large and small, accepted the key
outcomes around AI safety in Bletchley, suggesting that there are areas
where norms can be adopted, even if we disagree on other important
arenas of AI. Having the PRC endorse a declaration where protection of
human rights and transparency around AI are explicitly stated helps
align international norms with U.S. national interests.
Question. How is the Administration viewing the need to maintain a
dialogue with the PRC on avoiding AI's most significant risks while
also pushing back against the PRC's digital authoritarianism and its
efforts to influence global norms on AI?
Answer. Channels of communication are important to protect against
both expected and unexpected risks which may arise from the use or
deployment of emerging technologies, such as AI. Ensuring that safe and
rights-respecting AI is adopted, deployed, and used globally includes
discussions on AI at the UN and other global fora, which will
necessarily require interaction and conversations with the PRC. In
multilateral bodies across the United Nations, we are actively working
with like-minded partners at the International Telecommunications Union
(ITU) and the United Nations Educational, Science and Cultural
Organization (UNESCO) to stem the diffusion of critical technologies
that enable authoritarian misuse, to include taking tailored actions if
those governments seek to use these technologies to jeopardize our
national security.
Question. The PRC is ahead of the rest of the world when it comes
to establishing AI regulations with clear anti-compliance
repercussions, including rules requiring that AI generated content by
PRC companies is compliant with PRC law. How do you assess the PRC's
domestic regulatory actions on AI will impact their AI competitiveness
and innovation? How are other capitals struggle with similar questions,
how do you assess they are responding to the PRC model?
Answer. The PRC's regulations reflect a desire to increase control
over the outputs of AI systems to align with ``core socialist values,''
and potentially to use these systems for state-driven purposes, such as
surveillance. It is not yet clear how the PRC's regulatory actions will
impact their AI competitiveness and innovation.
One manner in which the PRC appeals to other countries in this
domain is to tout ``access to AI technologies'' for developmental
purposes. While the actual technologies and use cases remain nebulous,
the simple promise of development-oriented AI can attract support. We
maintain active and consistent engagement with partners and allies on
discussions of AI technologies and work to advance efforts,
particularly with developing countries, to push back against PRC
influence efforts related these technologies.
Question. Global AI Equity: Across the globe, leveraging AI could
help advance progress in critical areas such as healthcare, education,
and addressing issues related to climate. At the same time, some have
voiced concerns about the risk that developing countries could be left
behind from the benefits that advanced AI tools can bring to bear.
India in particular has sought to highlight this issue through its
Chairmanship of the Global Partnership for AI. How is the Department
seeking the participation of a broader range of stakeholders and
governments to work towards equitable distribution of the benefits of
AI?
Answer. Engaging a broad range of stakeholders is a key component
of the United States' strategy to ensure a cohesive approach to
realizing our vision for the global use of AI. We are working globally,
including with developing countries, to identify and promote broadly
accepted norms, guardrails, and governance models for this technology.
We are working in bilateral, multilateral, and multi-stakeholder venues
to build international partnerships driven by shared values and a
shared commitment to the responsible use of AI. The Department seeks to
ensure that the developing world benefits from AI-based technologies
and does so through a rights-respecting, democratic governance system.
At the ``Artificial Intelligence (AI) for Accelerating Progress on
the Sustainable Development Goals (SDGs)'' event during the 78th UN
General Assembly, Secretary Blinken announced that the State Department
will work with Congress to provide $15 million to ``help more
governments leverage the power of AI to drive global good, focused
specifically on the Sustainable Development Goals.'' The Department
anticipates allocating at least $7.5 million in funds from the CDP-
managed Digital Connectivity and Cybersecurity Partnership (DCCP) over
the next 5 years to promote the responsible use and governance of AI
globally. This allocation would include programs such as ``AI
Connect,'' which is designed to strengthen the multi-stakeholder AI
ecosystem, promote trustworthy AI applications and the responsible use
of AI, and empower low and middle-income countries to more effectively
participate in global conversations on AI and deploy AI-based solutions
to public policy challenges.
Question. G7 Hiroshima Process: The United States and other G7
members released an ``International Code of Conduct,'' which provides
voluntary guidance for organizations developing the most advanced AI
systems. How does the Hiroshima Process play into the Administration's
broader goals of leading with partners on AI governance
internationally?
Answer. The development of the ``International Code of Conduct for
Organizations Developing Advanced AI Systems'' by the G7, as a part of
the G7 Hiroshima AI Process, is an important step in showing that the
G7 can meet the AI moment and that democratic systems can move quickly
and deliver results for our citizens. This effort has helped to
internationalize our own domestic efforts such as the U.S. voluntary
commitments. The Code of Conduct is a significant diplomatic
achievement that advances the United States' approach to AI governance
by leveraging a larger multilateral framework. The Code consists of 11
principles that are directly inspired by and reinforce the U.S.
Voluntary Commitments. It represents the most substantive international
effort yet on international governance, and a common approach among the
world's leading democracies, released by G7 Leaders ahead of the UK AI
Safety Summit, amidst renewed UN interest in AI, and before adoption of
the anticipated EU AI Act. Several leading U.S. companies have already
voiced their support, and companies from other G7 countries are moving
toward support.
Question. Noting that the international code of conduct is a set of
voluntary principals, does the Administration anticipate that the
Hiroshima Process will lead to more enforceable actions in the future?
Answer. It is important that the United States and like-minded
partners begin discussions around AI regulation and governance with
voluntary principles to ensure strong alignment and coordination with
the private sector and research community and to ensure regulation does
not stifle innovation. The Department believes that domestic
regulations, which may include enforceable actions based on a risk-
based framework, can inform international discussions. The President
has voiced support for bipartisan legislation along these lines.
Question. U.S. Trade Representative Decision on Digital Trade: USTR
recently announced that it would no longer pursue digital trade rules
on data flows, data localization, source code protection, and digital
non-discrimination. As you conduct digital diplomacy with other
nations, how does USTR's decision affect your ability to advance a
free, open, and interoperable internet? How does it impact our efforts
to coordinate with allies and partners on AI research and development?
Answer. The United States is committed to advancing a fair,
inclusive, and innovative digital economy, and we continue the United
States' long-standing support for the trusted free flow of data and an
open Internet with strong and effective protections for individuals'
privacy as well as measures to preserve governments' abilities to
enforce laws and advance policies in the public interest.
Question. U.S. Federal Data Privacy Laws: One of the many risks
posed by the rapid development and deployment of artificial
intelligence is the impact that this technology will have on our
privacy. However, this is by no means a new problem. As you know, the
United States has struggled for years with some of our closest European
partners over issues related to data privacy protections for cross-
border data flows. How does our lack of federal data privacy laws
impact our ability to promote cooperation with allies and partners on
artificial intelligence research and development?
Answer. A bipartisan and comprehensive national data protection law
will add further credibility to the United States' efforts to build
international partnerships around AI governance, norms, and values, and
the Biden administration has strongly supported Congressional adoption
of comprehensive data protection law. The United States has long-
standing practices around privacy, has often been at the vanguard of
such discussions in the OECD, APEC, and elsewhere, and has robust
privacy safeguards in several key sectors such as health and financial
data, as well as strong protections on government handling of personal
data. Nevertheless, the lack of a comprehensive national data
protection law remains a key challenge when the United States is asking
other countries to make commitments around data and digital economy
matters, seeking to cooperate on AI research and development with our
allies and partners, and advocate on U.S. principles of privacy and
effective data governance.
Question. What is your assessment of the need for federal data
privacy legislation in terms of both safeguarding American's privacy
while also promoting U.S. innovation and competitiveness on artificial
intelligence? Is it a necessary foundation for potential subsequent
regulatory actions on artificial intelligence?
Answer. Comprehensive national data privacy legislation would add
further credibility to the United States' diplomatic efforts to shape
the international environment around AI governance and on other,
related digital policy issues. Trust and transparency in data use is an
important foundation to support future risk-based regulations on AI.
President Biden has called on Congress to pass bipartisan data privacy
legislation to better protect Americans' privacy, including from the
risks posed by AI.
Question. How effective has the U.S.-EU data sharing agreement been
in facilitating data flows between the U.S. and Europe?
Answer. The U.S.-EU Data Privacy Framework is an important
instrument to support the digital trade flows that underpin the
transatlantic digital economy. The Framework underscores with the EU
our shared commitment to privacy, data protection, the rule of law, and
our collective security, as well as our mutual recognition of the
importance of transatlantic data flows to our respective citizens,
economies, and societies. Data flows are critical to the transatlantic
economic relationship and for U.S. companies from all sectors of the
economy. The Framework has been effective in restoring legal certainty
and predictability in transatlantic data flows.
Question. Addressing the Spread of AI-Generated Disinformation: At
the UK AI Safety Summit, Vice President Harris called on all nations to
support the development and implementation of international standards
to help verify authentic government-produced digital content through
techniques such as watermarking and labeling. How is the State
Department assisting with this initiative?
Answer. The Department of State is working to advance mechanisms to
increase global resilience against deceptive or harmful synthetic AI-
generated or manipulated media. We will be working closely with civil
society, private sector, academia, standard-setting technical bodies,
industry, international community, and others to harmonize effective
authentication and identification standards through a multistakeholder
process to inform the U.S. Government and international content
authentication and provenance practices.
The Department is supporting global partners and civil society to
inform and implement content authentication standards. Our affirmative
vision is reflected in both our engagements and foreign assistance. We
plan to provide a range of support to ensure the development and
implementation of these standards is inclusive. First, we are
supporting civil society organizations to engage directly with key
standards setting bodies and to inform the U.S. Government about
international content authentication and provenance practices. Second,
we intend to sponsor capacity building and training programs to support
G7 partners, journalists, and other stakeholders to implement best
practices including those involved in establishing and implementing
content authentication and provenance standards. Third, the Department
of State convenes international Technology Challenges to identify local
and regional technologies that could be applied to address issues
including digital content authentication and AI-generated content
detection.
Question. What are the most promising tools or global standards for
digital content provenance?
Answer. The Department of State's Global Engagement Center (GEC)
encourages the development of guidelines, tools, and practices for
digital content authentication and synthetic content detection
capabilities to enhance high-integrity information exchange. The GEC
has openly encouraged the adoption of tools and safeguards to increase
the integrity and transparency of the information environment abroad,
to include international outreach to encourage and support the
development and implementation of uniform content authentication and
provenance standards. A promising strategy leverages multiple
technologies for digital content authentication rather than a single
best practice, such as using a combination of watermarking, labeling,
authentication blockchain technologies, steganographic techniques,
content modification logs, and hardware verification.
GEC is engaged in outreach efforts to the private sector and
academia, to include members of the Content Authenticity Initiative, as
they identify and implement technologies for content authentication and
provenance of high-integrity content. These efforts include
understanding the various forums and standards being enacted or
deliberated and how the U.S. Government should be involved or otherwise
learn from those efforts to inform our own policies and regulations.
Further, the GEC facilitates the use of a wide range of
technologies and techniques to reduce the malign use of AI by sharing
expertise among U.S. federal departments and agencies, and
international partners. The GEC executes a technology testbed to
rapidly test emerging technology applications for digital content
authentication and synthetic content detection measures. The GEC has
tested technology to authenticate foreign digital content and track its
provenance abroad and was an early tester of processes to label digital
content through watermarking.
______
Responses of Dr. Matthew Graviss to Questions
Submitted by Senator Benjamin L. Cardin
Question. Department Adoption of AI Tools: Data analytics and
machine learning tools, if deployed correctly and responsibly, present
an opportunity to inform and accelerate decision-making, to create
efficiencies to reduce the number of person-hours required to complete
complex tasks, to broaden the bandwidth to pursue other critical
missions, and to identify key areas of improvement.
Could you describe specific ways in which the Department has
integrated AI and machine learning tools into its work? How have they
enhanced the Department's ability to pursue its mission, particularly
as it relates to challenges like strategic competition and climate
change?
Answer. AI has progressively improved the Department's ability to
meet its mission. For example, the Department included a data scientist
in its COP27 delegation, allowing negotiators to have a decision-
advantage by modeling the impacts of proposals in real time. These same
climate-focused data scientists streamlined the creation of climate
negotiator briefing materials about 190 countries from 9 hours to 9
seconds; and, recently explored AI use in the robust collection of
information for climate evaluations.
Question. How are these tools and capabilities being coordinated
across the Department from bureau to bureau?
Answer. The Department has robust data governance policies and
coordination mechanisms at all organizational levels. As the
Department's Chief Data and AI Officer, I chair the Enterprise Data &
AI Council (EDAC), which sets the strategic direction for data and AI
utilization Department-wide. Subordinate to the EDAC is a robust Data
Governance Network, which includes Bureau-level Chief Data Officers
(BCDOs) and other data and AI practitioners across the Department to
coordinate tools, capabilities, and cross-bureau strategic initiatives.
Question. In an era of strategic competition, why is it important
that the Department of State embrace data analytics and machine
learning tools?
Answer. The United States is in a race for information and decision
advantage against the PRC, Russia, and other competitors. In a world of
information overload, data analytics, machine learning, and AI are
increasingly important tools for accelerating ``speed-to-insight,''
finding needles in haystacks, and providing national security
leadership with actionable information through analytical tools and
tradecraft--quantitative and qualitative alike. For example, we intend
to pursue responsible generative AI to make better, systematic use of
the information in our over 60 million diplomatic cables.
Question. While the State Department seeks to cooperate with
partners at an international level on responsible AI principles, it is
all the more critical that the Department has its own house in order in
terms of responsible AI use.
What oversight mechanisms does the Department have in place to
ensure that these tools are deployed safely and responsibly?
Answer. As the chair of the Enterprise Data and AI Council, I have
designated the Department's Responsible AI Official to chair the
offshoot AI Steering Committee, where technical experts in data
science, records management and privacy, cybersecurity, legal affairs,
and procurement address tactical questions of safe and responsible AI
deployment in accordance with E.O.s 14110, 13960, and 13859. These
groups oversaw development of the Enterprise AI Strategy (EAIS), which
articulates the responsible deployment of safe, secure, and trustworthy
AI.
Question. Department of State Enterprise AI Strategy: The
Department recently released its first ever ``Enterprise AI Strategy.''
Could you describe the vision that this strategy sets out for the
Department's future use of AI?
Answer. The vision of the Enterprise AI Strategy is that the
Department of State will responsibly and securely harness the full
capabilities of trustworthy artificial intelligence to advance United
States diplomacy and shape the future of statecraft. We will execute
this vision through four goals focused on secure infrastructure, an AI-
ready workforce, high-quality data assets, and responsible innovation.
The successful realization of this vision will depend on Department-
wide collaboration through the Enterprise Data and AI Council (EDAC)
and the Data and AI Campaigns it oversees.
Question. How does this new strategy shape or enhance the way the
Department's existing AI guidelines?
Answer. The Department's existing AI guidelines, outlined in 20 FAM
201.1, lay out a framework within which the Department's workforce can
use AI responsibly and largely aligns with E.O. 13960. The Enterprise
AI Strategy augments this existing policy framework by identifying
opportunities to embed AI into existing Department operational and
policy processes to improve efficiencies and reduce workload burden.
Additionally, the Strategy complements this policy framework by
initiating Department-wide AI maturity processes aligned to E.O. 14110.
Question. AI Workforce: The Administration's Executive Order has
placed an emphasis on the need to boost the Federal Government's
ability to hire individuals with the necessary skills to not only find
ways to leverage AI, but also to ensure that the adoption of AI is safe
and responsible.
What types of skillsets and backgrounds will be needed for the
Department to ensure that these tools are adopted responsibly and
ethically?
Answer. The Department has deployed data and AI modules into
mandatory tradecraft and DCM onboarding, in addition to its existing
slate of elective data and AI courses. Last year the Department
initiated the Bureau Chief Data Officer program, places senior data/AI
leaders in all bureaus alongside bureau leadership to responsibly
infuse data and AI into decision making processes. Additionally, the
Department has onboarded nearly five dozen job series 1560 data
scientists and U.S.D.C. fellows and have created a Locally Employed
Staff Data Scientist position description for all posts.
Question. What obstacles does the Department face in filling any
hiring gaps?
Answer. The Department strives to be a model employer and a home to
top-tier talent by positioning itself to rapidly onboard data and AI
practitioners that will strengthen our policy decisions and operational
processes. The Federal Government is generally at a disadvantage versus
private sector counterparts in terms of hiring. We are encouraged by
recent changes allowing for direct hire authority on relevant job
series. We are working to improve the Department's hiring
competitiveness through additional avenues such as specialized
incentive pay.
Question. In September, the Office of Personnel Management
announced a direct hire authority for federal agencies for data
scientist positions.
How does this authority impact the Department's ability to meet its
AI workforce needs? What else should this committee consider to ensure
that the Department is able to expand the use of data and AI to meet
its mission?
Answer. Direct hire authority will improve the Department's ability
to be a competitive employer in the marketplace by expediting the
hiring of data science and AI talent. The Department is seeking to
ensure its workforce can be compensated at a comparable rate to private
sector counterparts. Department organizational units require
specialized positions, like Bureau Chief Data Officers, who are
positioned at the bureau level to manage and govern data assets so that
data scientists can analyze data and leverage AI techniques to create
evidence for decision-makers.
Question. Administration EO and Digital Content Provenance: The
Administration's recent EO on artificial intelligence requires the
Department of State and other federal agencies to issue guidance for
authenticating the digital content they produce or publish.
What role will your office play in addressing the risks posed by
AI-generated content as outlined in the executive order?
Answer. As the Chief Data and AI Officer for the Department, my
office plays a central role in creating the governance framework around
responsible AI and any associated risk. The Responsible AI Official
(RAIO), who reports to me, is the Department's lead on AI risk and
chairs the AI Steering Committee (AISC)--a group of technical experts
that are charged with mitigating AI's various risks. As my office leads
the implementation of the Enterprise AI Strategy, the RAIO and AISC
will assemble policies and procedures for safe, secure, and ethical AI
use.
Question. How is your office coordinating internally at the
Department, including with offices like the Global Engagement Center,
to identify and address foreign disinformation operations enabled by
generative AI?
Answer. The Global Engagement Center leads the Department's efforts
to identify and address foreign disinformation and propaganda, is an
active member of the State Department's Enterprise Data and AI Council
(EDAC) and has shared tactics and findings in this space at EDAC
meetings. Additionally, the EDAC's current AI for Public Diplomacy (PD)
Campaign is sponsored by the Under Secretary for Public Diplomacy and
Public Affairs (R) and coordinates data and AI efforts across the R
Family (of which GEC is a part) and with PD practitioners at post.
Question. Does the Department of State intend to label and
authenticate the digital content posted on its public website?
Answer. E.O. 14110 outlines steps that federal departments and
agencies will take to make recommendations and issue guidance related
to labeling and authenticating digital content they create and share
with the public. As such, the Department, led by the Responsible AI
Official and the AI Steering Committee, will take steps to develop a
plan for such labels and authenticating measures, consistent with any
recommendations or guidance issued pursuant to the E.O. We want to make
sure the public will be able to trust when they see content from a U.S.
Government source.
Question. Are you aware of any existing tools or global standards
for digital content provenance?
Answer. We are aware of both standards and techniques used by the
public and private sector globally to certify digital content. It is
imperative that the Department, led through the Enterprise Data and AI
Council, work with digital creators in bureaus, offices, and posts and
across the interagency to leverage an appropriate metadata standard
that ensures content shared by the Department can continue to be
trustworthy.
______
Responses of Mr. Nathaniel Fick to Questions
Submitted by Senator Christopher A. Coons
Question. USTR recently announced that it would no longer seek to
advance digital trade rules on data flows, data localization, source
code protection, and digital non-discrimination. Does this decision
reflect an interagency consensus to abandon these longstanding
priorities that have been consistent across Administrations of both
parties?
Answer. We continue the United States' long-standing support for
the trusted free flow of data and an open Internet. We are committed to
advancing a fair, inclusive, and innovative digital economy, with
strong and effective protections for individuals' privacy as well as
measures to preserve governments' abilities to enforce laws and advance
policies in the public interest.
In the coming weeks and months, we will step up our engagement with
stakeholders--including both large and small companies in the
technology and other data-intensive sectors as well as privacy, safety,
labor, and human rights advocates--and work with our partners to chart
a path forward that appropriately balances these objectives. We will
continue to engage vigorously on the WTO's Joint Statement Initiative
on E-Commerce. We also look forward to working with Members of Congress
to support the legislative process on privacy, competition, and digital
regulation.
Question. If not, will you work with partners across the
interagency to arrive at a new consensus position that supports
American jobs, economic growth, and global competitiveness?
Answer. We are committed continued engagement with stakeholders on
this topic. We know these are complex issues, but we are bringing a
whole-of-government effort to find solutions and stepping up our
extensive engagement to chart a path forward. The Department is
committed to work across the interagency on these issues to support
American jobs, economic growth, and global competitiveness.
______
Responses of Mr. Nathaniel Fick to Questions
Submitted by Senator Brian Schatz
Question. Unregulated AI-generated content could destabilize
societies around the world. Although many other countries, including
the G7, the EU, even the People's Republic of China, have implemented,
or are pursuing AI labeling regulations, the United States has not.
Would U.S. diplomatic efforts be bolstered if the United States legally
required companies to label AI-generated content?
Answer. Since taking office, President Biden, Vice President
Harris, and the entire Biden-Harris administration have moved with
urgency to seize the tremendous promise and manage the risks posed by
Artificial Intelligence (AI) and to protect Americans' rights and
safety. As part of this commitment, President Biden convened seven
leading AI companies at the White House on July 21 to announce that the
Biden-Harris administration secured voluntary commitments from these
companies to help move toward safe, secure, and transparent development
of AI technology. As a result, several companies have committed
themselves to developing and implementing mechanisms to increase
transparency, such as content authentication and provenance that will
enable informing the public about the origins of their content, such as
labeling. At the same time, under the Executive Order issued on October
30, the U.S. Government committed to developing such mechanisms for its
own content. Creating effective mechanisms remains a significant
technical challenge that governments and companies around the world
must still address, but these initial efforts will bolster our ability
to coordinate with allies and partners around common, technically
feasible approaches to content authentication and provenance.
Question. How are you prioritizing AI labeling regulations in your
diplomatic discussions?
Answer. The United States is deeply engaged in discussions around
AI labeling in several fora with allies and partners, most prominently
the G7 and the OECD. In these fora the Department of State is working
to advance mechanisms and strategies to increase global resilience
against deceptive or harmful synthetic AI-generated digital content.
Question. Governments around the world are cutting off internet
access to suppress critical comments and remain in power. AI algorithms
can make this practice even more effective. What tools do we have to
discourage partners from cracking-down on internet freedom--whether by
using AI or other methods?
Answer. We are concerned by increasing efforts from governments to
control Internet access and content--including through the blunt use of
Internet shutdowns to suppress freedom of expression online--to stifle
dissent, quell protests, curb criticisms of the government, or limit
information leading up to and during elections. The U.S. Government has
made Internet freedom a major foreign policy priority for more than a
decade. The Department uses several tools to protect human rights
online, including raising our concerns bilaterally, building and
participating in multilateral and multi-stakeholder coalitions,
messaging our concerns through public diplomacy efforts, and providing
venture-capital-like grants to leading technologists and NGOs for
capacity building.
Regarding algorithms, we recognize that while AI can be immensely
helpful to technology companies in flagging content that may violate
their terms of use, it can also be used to suppress critical discourse.
We strongly discourage government intervention on content regulation
and instead encourage all governments to respect the independence of
technology platforms in developing and implementing content moderation
policies as per their terms of service.
Question. Can you provide examples of where those tools have been
successful and assess where the biggest challenges remain?
Answer. The Department works together with our country desk
officers, relevant bureaus, and U.S. embassies to monitor and respond
to laws, policies, and activities that impact human rights online, and
works to elevate concerns through appropriate channels, both publicly
and bilaterally. We also engage through international organizations and
UN bodies, such as the OECD and the G7, where we strive to develop
consensus around Internet freedom issues that we hope will establish
strong foundational norms. We continue to play a leadership role in the
Freedom Online Coalition (FOC), a group of 38 like-minded governments
committed to advancing human rights online. The FOC has been
instrumental in building cross-regional coalitions to support Internet
freedom, addressing related challenges in multilateral venues, and
creating a space for collaboration with civil society and business
around related emerging and critical issues, including AI.
Question. Besides U.S.-led efforts, which multilateral and
international organizations should we be prioritizing to help promote
internet freedom in the face of AI?
Answer. In multilateral bodies across the United Nations, such as
in the International Telecommunications Union (ITU) and the United
Nations Educational, Science and Cultural Organization (UNESCO), we are
actively working with like-minded partners to stem the diffusion of
critical technologies that enable authoritarian misuse, to include
taking narrowly-focused actions if those governments seek to use these
technologies to jeopardize our national security. The United States
will continue to lead and support global norm setting for the use of AI
to ensure the adoption and deployment of safe and values-based
standards.
Question. What guardrails must be put in place to set AI
development on a path that bolsters democracy, not undermines it?
Answer. All actors developing, deploying, and using AI can and
should be implementing measures throughout the lifecycle of AI systems
to manage the risks this technology can pose to democracy as well as
support civil and human rights. Such measures include conducting impact
assessments, sufficiently testing AI systems in appropriate real-world
contexts, identifying and mitigating factors contributing to
algorithmic discrimination or other disparate impacts on specific
populations, and conducting ongoing monitoring of deployed AI systems.
Other measures include creating options for ongoing feedback, repeal,
and recourse by users and other impacted stakeholders; training AI
operators on risk identification, assessment, prevention, and
mitigation; prioritizing cybersecurity by design in AI systems; as well
as implementing transparency measures such as watermarking synthetic
content or disclosing when decisions are made by AI systems.
Question. How do you weigh the tradeoffs of including the PRC,
which staunchly opposes freedom of expression, in international
negotiations over AI?
Answer. AI will impact nearly all facets of government and society
around the world, and a multistakeholder and international approach is
necessary to mitigate potential risks to people's rights and safety. We
also recognize that international cooperation is essential to building
a safe, secure, and trustworthy AI future. A multistakeholder approach
is critical to build international consensus around safe, secure,
responsible, ethical, trusted, and transparent use of AI technology
that protects, among other principles, the freedom of expression.
Through this approach, the United States engages with many
countries on AI issues, including those active in the development of AI
technologies, such as the PRC. As part of these conversations, the
United States and the international community deploys strategic, long-
term, and persistent diplomacy to continue to protect and promote
freedom of expression and human rights. Overall, ensuring that safe and
rights-respecting AI is adopted, deployed, and used globally, includes
discussions on AI at the UN and other global fora that will necessarily
require interaction and conversations with the PRC.
Letter to President Biden, Dated November 30, 2023,
Concerning the Decision of the USTR To Stop Supporting
Key Commitments in the E-Commerce Negotiations at the WTO
Submitted by Senator Chris Van Hollen
[GRAPHICS NOT AVAILABLE IN TIFF FORMAT]
[all]